hash stringlengths 32 32 | doc_id stringlengths 7 13 | section stringlengths 3 121 | content stringlengths 0 2.2M |
|---|---|---|---|
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 10.3.7 Performance evaluation tools | Performance Evaluation Tools rate processed metric values according to a set of rules. The purpose is to automatically produce a verdict about measured performance of the SUT. For reliability metrics Performance Evaluation Tools will process captured performance data to identify trends or irregular behaviour that could endanger the service production. For efficiency metrics both trend spotting and rule based checks applies. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 10.3.8 Performance presentation tools | Performance Presentation Tools transform measured performance into graphs and other presentation formats. The purpose is to improve and enhance interpretation of measured performance. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 11 Performance test specifications | |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 11.1 Elements of performance test specifications | Specifications on a performance test include the following elements: 1) test objectives; 2) test conditions; 3) test configurations; 4) test data specifications; and 5) test evaluation specifications. Performance test specifications are translated into performance test configurations, i.e. configurations of Test System, System Under Test and Test Bed infra structure to enable collection of performance data. Performance test specification elements Test objectives Test conditions Test configurations Test data specifications Test evaluation specifications Figure 21: Elements of performance test specifications |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 11.2 Test objectives | The Test objectives of a performance test state the purposes of the test, i.e. what will be achieved by running the test. ETSI ETSI TR 101 577 V1.1.1 (2011-12) 66 |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 11.3 Test conditions | The Test conditions of a performance test include: 1) test specification prerequisites; 2) test execution pre-conditions; 3) operational measurement conditions; and 4) test execution post-conditions. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 11.3.1 Test specification prerequisites | With test specification prerequisites we mean output from other performance tests that is required as input to a performance test specification. A consequence is that such performance tests have to be executed before specifying the intended performance test. Performance tests of stability and availability characteristics is an example of test cases that require the output from other performance tests as input to the test specifications. A performance test of stability and availability characteristics usually runs for days or even weeks at 80 % of a system's measured maximum throughput level. A prerequisite to specify such a test is information about what is 80 % of a system's measured throughput level. This information is obtained in a performance test of the system's Sustained throughput capacity. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 11.3.2 Test Execution Pre-conditions | A set of Test Bed and the SUT conditions expected to be met before start of Performance Test execution. These conditions are referred to as Test Pre-conditions. The Pre-conditions also apply after the initial test steps (the warm-up phase when simulated users get activated and the system is prepared to receive service requests) have completed. Pre-conditions are usually stated for the performance test bed as well as the tested application. Test bed pre-conditions Examples of test bed pre-conditions are: 1) exclusive access to the performance test bed; 2) enough physical resources for execution of a performance test case, such as disk space; and 3) correct configuration of the test bed components, such as test tool equipment connected to all open SUT interfaces. Application pre-conditions Examples of application pre-conditions are: 1) ensure that all simulated entities are in correct state to run the test, such as all simulated entities are successfully registered and accepted by the system. (Not always required); 2) ensure that common application resources, such as databases, contain expected data. (Not always required); 3) ensure that required load is applied on the SUT; and 4) ensure that the SUT can process requested application services. ETSI ETSI TR 101 577 V1.1.1 (2011-12) 67 |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 11.3.3 Operational Measurement conditions | Operational measurement conditions state conditions that apply when measurement data are captured during an on-going performance test. Operational measurement conditions may also state under what circumstances a performance test can be stopped. There are two types of Measurement conditions: 1) requested measurement conditions; and 2) actual measurement conditions. Requested measurement conditions Requested measurement conditions are stated requirements on SUT and Test Bed (Test Tool) conditions for capturing performance data. Requested measurement conditions are a part of the performance test specification. Requested measurement conditions can be external or internal. External measurement conditions External measurement conditions describe what a tested system (SUT) is to be exposed to during a performance test. External conditions include types of service requests, volumes of service requests (traffic rates), duration of service request volumes, and volumes of simulated entities or users requesting services. Internal measurement conditions Internal measurement conditions describe expected situations inside a tested system during a performance test, such as resource usage levels of CPU, memory, etc. Actual measurement conditions Actual measurement conditions are recorded conditions that applied when performance data were captured. The purpose of Actual measurement conditions is to validate recorded performance data. Actual measurement conditions include metrics such as load deviations - the differences between intended load and actual load during a performance test. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 11.3.4 Test Execution Post-conditions | A set of Post-conditions expected to be met after execution of the performance test has completed and before the performance test is regarded as completed. Examples of post-conditions activities are: 1) all system resources reserved during test execution are released. such as all sessions, subscriptions, or other resources related to pending services are returned; and 2) central resources on the test bed are reset, such as used databases. The purpose of post-conditions is to bring the test environment to a well defined initial state for the next test. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 11.4 Test configurations | A set of specifications that apply to the Test System and the System Under Test and enable execution of the performance test. Performance test configurations include: • workload specifications; • test bed specifications; and • data collection specifications. A performance test configuration covers in most cases more than one performance test cases. ETSI ETSI TR 101 577 V1.1.1 (2011-12) 68 |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 11.4.1 Workload specifications | Workload specifications describe what a System Under Test is expected to handle or process during a performance test. Workload is described more in detail in clause 12. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 11.4.2 Test bed specifications | The Test bed specifications describe the test bed configuration of equipment for different services in a performance test. The Test bed specifications for interfaces between the test tools and the SUT describe: • IP addresses and listening ports of the SUT for different services; • used network layer protocols such as IPv4 and/or IPv6; • used transport layer protocols such as TCP, UDP, SCTP etc; • used application layer protocols such as HTTP, SOAP, SIP, Radius, Diameter, DHCP etc; • security settings such as IPsec or HTTPS; and • timeout settings for service requests. Other Test bed specifications describe: • the hardware configurations of servers in the SUT; • the number of servers and load balancing equipment in the SUT; • the Test bed equipment interconnecting SUT and Test tools; and • requested versions of all software involved in a performance test. The purpose of the Test bed specifications is to document the test environment such that a performance test can be repeated identically at any point in time. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 11.4.3 Data collection specifications | Data collection specifications contain specifications of performance data that should be captured. Data collection specifications include: 1) specifications of internal performance data collection; 2) specifications of external performance data collection; and 3) specifications of performance recording details. Specifications of internal performance data collection Internal performance data collection specifications include: • specifications of selected performance variables, such as CPU usage, memory usage, or queues; • recording location of selected performance variables, such as per server or for a specified process group; and • the frequency of recording internal performance data inside the SUT. ETSI ETSI TR 101 577 V1.1.1 (2011-12) 69 Specifications of external performance data collection External performance data collection specifications include: • specifications of selected performance variables for requested services and related responses; and • specifications of selected performance variables for actual measurement conditions. Specifications of performance recording details The performance recording details include configuration parameters such as: • the resolution of recorded performance data, such as seconds, milliseconds, or microseconds for response time; and • the frequency of saving captured performance data on disk, i.e. sample time for recording performance data. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 11.5 Test Data Specifications | Test data specifications is a set of specifications to enable Authentication of simulated users, authorization of service requests, customization of service requests, and evaluation of test results. Test data specifications contain three parts: • test data for service requests; • test data for SUT operability; and • test data for performance evaluation. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 11.5.1 Test Data for service requests | Test data for service requests is a set of parameters values for every simulated user that is used to make every service request from every simulated user unique. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 11.5.2 Test Data for SUT operability | Test data for SUT operability is set of unique parameter values for every simulated user that corresponds to stored information about the users in the SUT's databases, such as identities, phone numbers, account numbers, etc. The Test Data for SUT operability is required to authenticate and authorize service requests from simulated users during performance tests, i.e. a prerequisite for SUT operability. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 11.5.3 Test Data for performance evaluation | Test data for evaluation of performance measurement is a set of evaluation rules and expected measurement values. Test Data for performance evaluation of regression tests contain the performance measurement results from some previous execution of the performance test together with the evaluation criteria. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 11.6 Test evaluation specifications | Test evaluation specifications are a set of rules and settings for evaluation and rating of performance test results. Test evaluation specifications will for instance specify ranges for performance measurement result when an appropriate verdict such as Excellent, Good, Acceptable, Poor, and Bad, etc. should be applied. ETSI ETSI TR 101 577 V1.1.1 (2011-12) 70 |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 12 Workload concepts | Workload is a term for what a System Under Test is expected to handle or process totally during a performance test. A Workload has three components: • workload content; • workload volume; and • workload time distribution. A work load can also define what load a SUT is exposed to at a given point of time during the test. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 12.1 Workload set or Traffic set | A performance test may contain several workload specifications, where each workload specifies a group of simulated users exposing the SUT to a specific set of service requests with a specified intensity. A set of Workload configurations is also called a Traffic Set. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 12.2 Workload content | The Workload content describes the services that will be requested from the SUT during a performance test. The Workload content contains two parts: 1) a User Session Scenario or a Requested Service Profile; and 2) a set of Service Scenarios. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 12.2.1 User Session Scenarios | A User Session Scenario describes the Workload content as seen from the requesting side (the TS). A User Session Scenario contains a list of services requested during a user session and in which order the services are requested, i.e. the flow of service requests in a user session. The flow also contains alternative paths depending on the outcome of a service request. A User Session Specification usually has some exits for emergency situations too. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 12.2.2 Requested Service Profile | A Requested Service Profile describes the Workload content as seen from the receiving side (the SUT). The Requested Service Profile contains a list of requested services. Each service request has a specification of how frequently it should be generated expressed as a percentage of all requests. The sum of percentage values for all service requests is 100. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 12.2.3 Service scenarios | Service Scenarios are specifications of individual service requests and include topics such as: • how requests are sent to the SUT; • how response messages from the SUT are validated; • how errors reported from the test bed, such as timeout or disconnects, are handled; • how the outcome of a service request is reported back to the User session; and • building a service request with requested content and formatted with user specific information. ETSI ETSI TR 101 577 V1.1.1 (2011-12) 71 |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 12.3 Workload volume | The Workload volume is a description of the amount of services (described in the Workload content) that is requested during a performance test. Workload volume and workload time distribution are also referred to as Load characteristics for a performance test. The Workload volume is specified differently depending on the type of load generation (see Load concepts below). At User session driven load Workload volume is defined as a set of User session load steps, where each load step specifies a number of concurrently simulated users and duration. At Traffic rate driven load Workload volume is defined as a set of Traffic rate load steps, where each load step specifies a traffic rate and duration. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 12.4 Load concepts | Depending on the design of a test tool there are two concepts for generating load: 1) User session driven load 2) Traffic rate driven load |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 12.4.1 User session driven load | User session driven load is based on traffic generated by a number of simulated users, where the rate of service requests from each user is controlled by think-time delays between consecutive service requests. The total load on the system is in this case determined by the number of concurrently active user sessions. In order to increase the system load more simulated user sessions have to start. The number of concurrently active user sessions during a performance test is controlled in a load script. In some cases load can be manually controlled during a test. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 12.4.2 Traffic rate driven load | Traffic rate driven load is based on traffic controlled by a central load control function in the test tool keeping track of when a user session is instructed to send a request. The traffic rate specification is independent of the number of simulated entities. The load control function executes a load script telling what traffic rate should be applied in every moment. Each specified traffic rate in the load script has a duration time. The total performance test duration is set by the sum of all specified duration times. Transition time between two traffic rates is instantaneous. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 12.5 Workload time distribution | The Workload time distribution is a description of how the amount of requested services (the Workload volume) is distributed over time during a performance test. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 12.5.1 Load profiles | A Load profile is the set of load conditions defined in a load script. |
b1381823cf46aa778d0671bd8826a4bf | 101 577 | 12.5.2 Load patterns | The load on the SUT can follow several types of patterns such as: • constant load; • peak load; ETSI ETSI TR 101 577 V1.1.1 (2011-12) 72 • saw tooth load; • statistical load; and • stepwise increased load. Constant load A load pattern where the SUT is exposed to a fixed rate of service requests per time unit. Constant load is commonly used in performance tests of stability and availability characteristics. Peak load A load pattern where the SUT is exposed to a repeated sequence of short periods of very high load (peaks) followed by longer periods of low load. Peak load is commonly used in performance tests of robustness characteristics. Saw tooth load A load pattern where the SUT is repeatedly exposed to a sequence of increasing and decreasing load. Saw tooth load is usually commonly in performance tests of robustness characteristics. Statistical load A load pattern where the SUT is exposed to load according to a statistical model for arrival of service requests, such as a Poisson or F distribution or Erlang. Statistical load is usually commonly in simulations of service production in system delivery tests. Stepwise increased load A load pattern where the SUT is exposed to a sequence of fixed load increases. Stepwise increased load is commonly used in performance tests of capacity characteristics. Stepwise increased load is sometimes called staged load. ETSI ETSI TR 101 577 V1.1.1 (2011-12) 73 History Document history V1.1.1 December 2011 Publication |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 1 Scope | The present document provides implementation and usage guidelines for higher-layer functions in DVB-RCS2 interactive satellite networks, which is defined in [i.2]. The lower-layer specification and implementation guidelines for DVB-RCS2 networks are presented in [i.3] and [i.4], respectively. The present document covers on transparent star, regenerative mesh, and transparent mesh overlay network topologies. The recommendations and examples provided in the present document are informative. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 2 References | References are either specific (identified by date of publication and/or edition number or version number) or non-specific. For specific references, only the cited version applies. For non-specific references, the latest version of the referenced document (including any amendments) applies. Referenced documents which are not found to be publicly available in the expected location might be found at http://docbox.etsi.org/Reference. NOTE: While any hyperlinks included in this clause were valid at the time of publication ETSI cannot guarantee their long term validity. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 2.1 Normative references | Not applicable. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 2.2 Informative references | The following referenced documents are not necessary for the application of the present document but they assist the user with regard to a particular subject area. [i.1] ETSI TS 101 545-3: "Digital Video Broadcasting (DVB); Second Generation DVB Interactive Satellite System (DVB-RCS2); Part 3: Higher Layers Satellite Specification". [i.2] ETSI TS 101 545-1: "Digital Video Broadcasting (DVB); Second Generation DVB Interactive Satellite System (DVB-RCS2); Part 1: Overview and System Level specification". [i.3] ETSI EN 301 545-2 V1.1.1 (2012-01): "Digital Video Broadcasting (DVB); Second Generation DVB Interactive Satellite System (DVB-RCS2); Part 2: Lower Layers for Satellite standard". [i.4] ETSI TR 101 545-4: "Digital Video Broadcasting (DVB); Second Generation DVB Interactive Satellite System (DVB-RCS2); Part 4: Guidelines for Implementation and Use of EN 301 545-2". [i.5] IETF RFC 2328: "OSPF Version 2". [i.6] IETF RFC 2453: "RIP Version 2". [i.7] IETF RFC 5340: "OSPF for IPv6". [i.8] IETF RFC 4271: "A Border Gateway Protocol 4 (BGP-4)". [i.9] IETF RFC 5880: "Bidirectional Forwarding Detection (BFD)". [i.10] IETF RFC 5881: "Bidirectional Forwarding Detection (BFD) for IPv4 and IPv6 (Single Hop)". [i.11] IETF RFC 1112: "Host Extensions for IP Multicasting". [i.12] IETF RFC 2365: "Administratively Scoped IP Multicast". [i.13] IETF RFC 2236: "Internet Group Management Protocol, Version 2". ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 10 [i.14] IETF RFC 3376: "Internet Group Management Protocol, Version 3". [i.15] IETF RFC 4606: "Generalized Multi-Protocol Label Switching (GMPLS) Extensions for Synchronous Optical Network (SONET) and Synchronous Digital Hierarchy (SDH) Control". [i.16] IETF RFC 3810: "Multicast Listener Discovery Version 2 (MLDv2) for IPv6". [i.17] IETF RFC 4601: "Protocol Independent Multicast - Sparse Mode (PIM-SM): Protocol Specification (Revised)". [i.18] IETF RFC 4605: "Internet Group Management Protocol (IGMP) / Multicast Listener Discovery (MLD)-Based Multicast Forwarding ("IGMP/MLD Proxying")". [i.19] IETF RFC 4541: "Considerations for Internet Group Management Protocol (IGMP) and Multicast Listener Discovery (MLD) Snooping Switches". [i.20] IETF RFC 3171: "IANA Guidelines for IPv4 Multicast Address Assignments". [i.21] IETF RFC 2474: "Definition of the Differentiated Services Field (DS Field) in the IPv4 and IPv6 Headers". [i.22] IETF RFC 2475: "An Architecture for Differentiated Services". [i.23] IETF RFC 3290: "An Informal Management Model for Diffserv Routers". [i.24] IETF RFC 3086: "Definition of Differentiated Services Per Domain Behaviors and Rules for their Specification". [i.25] IETF RFC 2753: "A Framework for Policy-based Admission Control". [i.26] IETF RFC 2698: "A Two Rate Three Color Marker". [i.27] IETF RFC 2697: "A Single Rate Three Color Marker". [i.28] IETF RFC 3246: "An Expedited Forwarding PHB (Per-Hop Behavior)". [i.29] IETF RFC 3247: "Supplemental Information for the New Definition of the EF PHB (Expedited Forwarding Per-Hop Behavior)". [i.30] IETF RFC 2597: "Assured Forwarding PHB Group". [i.31] IETF RFC 4594: "Configuration Guidelines for DiffServ Service Classes". [i.32] IETF RFC 3584: "Coexistence between Version 1, Version 2, and Version 3 of the Internet- standard Network Management Framework". [i.33] IETF RFC 3413: "Simple Network Management Protocol (SNMP) Applications". [i.34] IETF RFC 3415: "View-based Access Control Model (VACM) for the Simple Network Management Protocol (SNMP)". [i.35] IETF RFC 3411: "An Architecture for Describing Simple Network Management Protocol (SNMP) Management Frameworks". [i.36] IETF RFC 3918: "Methodology for IP Multicast Benchmarking". [i.37] IETF RFC 3412: "Message Processing and Dispatching for the Simple Network Management Protocol (SNMP)". [i.38] IETF RFC 3414: "User-based Security Model (USM) for version 3 of the Simple Network Management Protocol (SNMPv3)". [i.39] IETF RFC 5728: "The SatLabs Group DVB-RCS MIB". [i.40] ETSI TS 132 101: "Digital cellular telecommunications system (Phase 2+); Universal Mobile Telecommunications System (UMTS); LTE; Telecommunication management; Principles and high level requirements (3GPP TS 32.101)". ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 11 [i.41] ETSI TS 132 150: "Digital cellular telecommunications system (Phase 2+); Universal Mobile Telecommunications System (UMTS); LTE; Telecommunication management; Integration Reference Point (IRP) Concept and definitions (3GPP TS 32.150)". [i.42] 3GPP TS 42.435: "Telecommunication management; Performance measurement; eXtensible Markup Language (XML) file format definition". [i.43] ETSI TS 132 300: "Digital cellular telecommunications system (Phase 2+); Universal Mobile Telecommunications System (UMTS); LTE; Telecommunication management; Configuration Management (CM); Name convention for Managed Objects (3GPP TS 32.300)". [i.44] ETSI TS 132 405: "Digital cellular telecommunications system (Phase 2+); Universal Mobile Telecommunications System (UMTS); LTE; Telecommunication management; Performance Management (PM); Performance measurements; Universal Terrestrial Radio Access Network (UTRAN) (3GPP TS 32.405)". [i.45] 3GPP TS 22.228 V12.0.0 (2011-12): "Service requirements for the Internet Protocol (IP) Multimedia core network Subsystem (IMS), Stage 1". [i.46] 3GPP TS 23.203 V11.4.0 (2011-12): "Technical Specification Group Services and System Aspects; Policy and charging control architecture (Release 11)". [i.47] ETSI TS 132 240: "Digital cellular telecommunications system (Phase 2+); Universal Mobile Telecommunications System (UMTS); LTE; Telecommunication management; Charging management; Charging architecture and principles (3GPP TS 32.240)". [i.48] 3GPP TS 23.402 V11.0.0 (2011-09): "Technical Specification 3rd Generation Partnership Project; Technical Specification Group Services and System Aspects; Architecture enhancements for non- 3GPP accesses (Release 11)". [i.49] 3GPP TS 29.212 V11.2.0 (2011-09): "Technical Specification 3rd Generation Partnership Project; Technical Specification Group Core Network and Terminals; Policy and Charging Control (PCC) over Gx/Sd reference point (Release 11)". [i.50] 3GPP TS 29.213 V11.0.0 (2011-09): "Technical Specification 3rd Generation Partnership Project; Technical Specification Group Core Network and Terminals; Policy and Charging Control signalling flows and Quality of Service (QoS) parameter mapping (Release 11)". [i.51] IETF RFC 3588: "Diameter Base Protocol". [i.52] IETF RFC 5213: "Proxy Mobile IPv6". [i.53] IETF RFC 4301: "Security Architecture for the Internet Protocol". [i.54] IETF RFC 3522: "The Eifel Detection Algorithm for TCP". [i.55] IETF RFC 4015: "The Eifel Response Algorithm for TCP". [i.56] IETF RFC 5682: "Forward RTO-Recovery (F-RTO): An Algorithm for Detecting Spurious Retransmission Timeouts with TCP". [i.57] M. Fiedler, T. Hossfeld, and P. Tran-Gia: "A Generic Quantitative Relationship between Quality of Experience and Quality of Service" IEEE Network, vol. 24, no. 2, Apr. 2010, pp. 36-41. [i.58] H. Skinnemoen, A. Vermesan, A. Iuoras, G. Adams, and X. Lobao: "VoIP over DVB-RCS with QoS and bandwidth on demand" IEEE Wireless Communications, vol.12, no.5, pp. 46- 53, Oct. 2005. [i.59] IETF RFC 5681: "TCP Congestion Control". [i.60] IETF RFC 6298: "Computing TCP's Retransmission Timer". [i.61] N. Dukkipati, T. Refice, Y. Cheng, J. Chu, N. Sutin, A. Agarwal, T. Herbert, and J. Arvind: "An Argument for Increasing TCP's Initial Congestion Window", ACM SIGCOMM Computer Communications Review, vol. 40, pp. 27-33, July 2010. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 12 [i.62] J. Chu, N. Dukkipati, Y. Cheng, and M. Mathis: "Increasing TCP initial Window", Internet Draft, draft-hkchu-tcpm-initcwnd-01.txt, July 2010. [i.63] D. J. Wischik: "Short Messages", Philosophical Transactions of the Royal Society A, vol. 366, pp. 1941-1953, 2008. [i.64] IETF RFC 5690: "Adding Acknowledgement Congestion Control to TCP". [i.65] IETF RFC 3390: "Increasing TCP's Initial Window". [i.66] Y. Chen: "Seeding RTO with RTT sampled during three-way handshake", Internet Draft, draft- ycheng-tcpm-rtosynrtt.txt, IETF, June 2010. [i.67] Y. Chen, J. Chu, and A. Jain: "TCP Fast Open", Internet Draft, draft-cheng-tcpm-fastopen-00.txt, March 2011. [i.68] IETF RFC 793: "Transmission Control Protocol". [i.69] IETF RFC 1901: "Introduction to Community-based SNMPv2". [i.70] IETF RFC 1905: "Protocol Operations for Version 2 of the Simple Network Management Protocol (SNMPv2)". [i.71] IETF RFC 5882: "Generic Application of Bidirectional Forwarding Detection (BFD)". [i.72] IEEE 802.1pQ: "IEEE Standard for Local and metropolitan area networks--Media Access Control (MAC) Bridges and Virtual Bridged Local Area Networks". |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 3 Definitions, symbols and abbreviations | |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 3.1 Definitions | For the purposes of the present document, the terms and definitions given in [i.1] apply. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 3.2 Symbols | For the purposes of the present document, the symbols given in [i.1] apply. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 3.3 Abbreviations | For the purposes of the present document, the following abbreviations apply: 3GPP 3rd Generation Project Partnership 3WHS 3-Way HandShake AAA Authentication Authorization Accounting AAR Authentication Authorisation Request ABR Area Border Router AC Allocation Channel ACK ACKnowledgement ACM Adaptive Coding and Modulation ADC Application Detection and Control AF Assured Forwarding AH Authentication Header ALPDU Addressed Link Protocol Data Unit AR Address Resolution ARP Allocation and Retention Priority AS Autonomous System ATM Asynchronous Transfer Mode AVBDC Absolute Volume Based Dynamic Capacity ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 13 AVP Attribute Value Pair BA Behaviour Aggregate BBERF Bearer Binding and Event Reporting Function BCT Broadcast Configuration Table BDR Backup Designated Router BE Best Effort BER Bit Error Rate BFD Bidirectional Forwarding Detection BGP Border Gateway Protocol BSS Business Support System BW Bandwidth CA Connectivity Aggregate CAC Connection Admission Control CAN Connectivity Access Network CC Connectivity Channel or Capacity Category CCA Credit Control Answer CCR Credit Control Request CEA Capabilities Exchange Answer CER Capabilities Exchange Request CFI Canonical Format Indicator CLI Command Line Interface CM Configuration Management CNG Customer Network Gateway CNR Carrier to Noise Ratio COMSEC Communication Security CoS Class of Service CPN Customer Premises Network CPU Central Processor Unit CR Capacity Request CRA Constant Rate Assignment CS Class Selector CSCF Call Session Control Function CW Continuous Wave DA Dedicated Access DA-AC Dedicated Access Allocation Channel DAMA Demand Assignment Multiple Access DCP Dynamic Connectivity Protocol DF Default Forwarding DHCP Dynamic Host Control Protocol DN Distinguished Name DNS Domain Name Server DR Designate Router DS Differentiated Services DSCP Differentiated Services Code Point DUP DUPlicate DVB Digital Video Broadcast DVB-RCS2 Digital Video Broadcast Return Channel via Satellite 2nd generation DVB-S Digital Video Broadcast - Satellite ECN Explicit Congestion Notification EF Expedited Forwarding EIRP Effective Isotropically Radiated Power EM Elements Manager ESP Encapsulating Security Payload FCA Free Capacity Allocation FCAPS Fault, Configuration, Accounting, Performance, Security FCT2 Frame Composition Table 2nd generation FEC Forward Error Correction FIB Forwarding Information Base FIFO First In First Out FL Forward Link FPDU Frame Protocol Data Unit ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 14 FTP File Transfer Protocol GBR Guaranteed Bit Rate GPS Global Positioning System GRE Generic Routing Encapsulation GSE Generic Stream Encapsulation GSM Global System Mobile GW GateWay HID Hardware IDentifier HL Higher Layer HLID Higher Layers Initialisation Descriptor HLS Higher Layer Service HSS Home Subscriber Server HTTP Hyper-Text Transfer Protocol HTTPS Hyper-Text Transfer Protocol Secure HW Hardware IANA Internet Assigned Numbers Agency IB Installation Burst ICMP Internet Control Message Protocol IE Information Element IETF Internet Engineering Task Force IF Intermediate Frequency IFL Inter-Facility Link IGMP Internet Group Management Protocol IMS Internet Multimedia Subsystem IMSI International Mobile Subscriber Identity INID Interactive Network ID IP Internet Protocol IP-CAN Internet Protocol – Connectivity Access Network IPTV Internet Protocol TV IPv4 Internet Protocol version 4 IPv6 Internet Protocol version 6 IRP Interface Reference Point ISC IMS Service Control ISO International Standards Organisation ISP Internet Service Provider ITU-T International Telecommunication Union - Telecommunication IW Initial Window KPI Key Performance Indicator L2S Layer-2 Signalling LAN Local Area Network LB Link Behaviour LCD Local Configuration Datastore LDN Local Distinguished Name LI Link Interface LL Lower Layer LLS Lower Layer Service LMA Local Mobility Anchor LNB Low Noise Block LQC Link Quality Control LS Link Stream LSA Link State Advertisement LSB Least Significant Bit LSE Link Service Establishment LTE Long Term Evolution LW Loss Window MAC Medium Access Control MAC24 A 24-bit MAC address MAG Mobile Access Gateway MBR Maximum Bit Rate MC Mesh Controller MF Multi-field MFIB Multicast Forwarding Information Base ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 15 MF-TDMA Multi-Frequency Time Division Multiple Access MIB Management Information Base MLD Multicast Listener Discovery MMT Multicast Mapping Table 1st generation MMT2 Multicast Mapping Table 2nd generation MPEG Moving Pictures Expert Group MSB Most Significant Bit MSS Maximum Segment Size MTU Maximum Transmission Unit NAPT Network Address Port Translation NBMA Non-Broadcast Multiple Access NCC Network Control Centre NCR Network Clock Reference NE Network Element NGN Next Generation Networks NIT Network Information Table NLID Network Layer Information Descriptor NM Network Manager NMC Network Management Centre NMS Network Management System OAM Operations Administration Maintenance OAM&P Operations Administration Maintenance Provisioning OCS Online Charging System ODU OutDoor Unit OFCS Offline Charging System OID Object ID ONID Original Network ID OSPF Open Shortest Path First OSS Operations Support System OUI Organisationally Unique IDentifier OVN Operator Virtual Network PBA Proxy Binding Acknowledgement PBU Proxy Binding Update PCC Policy and Charging Control PCEF Policy Control and Charging Enforcement Function PCP Priority Code Point PCRF Policy Control and Charging Rules Function PDN Packet Data Network PDP Packet Data Protocol PDU Protocol Data Unit PEP Performance Enhancing Proxy or Policy Enforcement Point PHB Per-Hop Behaviour PID Packet IDentifier PIM Protocol Independent Multicast PIM-SM Protocol Independent Multicast - Sparse Mode PLMN Public Land Mobile Network PM Performance Management PMIP Proxy Mobile Internet Protocol PPDU Payload-adapted Protocol Data Unit QCI QoS Class Identifier QoS Quality of Service RA Random Access RAA Re-Authentication Answer RA-AC Random Access Allocation Channel RAR Re-Authentication Request RBDC Rate Based Dynamic Capacity RC Request Class RCS Return Channel via Satellite RCST Return Channel via Satellite Terminal RCST2 Return Channel via Satellite Terminal 2nd generation RF Radio Frequency ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 16 RFC Request For Comments RIB Routing Information Base RIP Routing Information Protocol RL Return Link RLE Return Link Encapsulation RMT RCS Map Table RNC Radio Network Controller RO Read Only RPF Reverse Path Forwarding RPLS Receiver Physical Layer Segment RRM Radio Resource Management RTD Round Trip Delay RTO Retransmission TimeOut RTT Round Trip Time RW Read Write Rx Receive/Receiver SA Service Aggregate SACK Selective ACKnowledgement SCADA Supervisory Control And Data Acquisition SCPC Single Channel Per Carrier SCTP Streaming Control Transport Protocol SDDP Software and Data Distribution Protocol SDF Service Data Flow SDP Session Description Protocol SDU Service Data Unit SI Service Information SIP Session Initiation Protocol SLA Service Level Agreement SMSS Sender Maximum Segment Size SNMP Simple Network Management Protocol SNO Satellite Network Operator SNR Signal to Noise Ratio SOHO Small Office Home Office SP Service Provider SPR Subscription Profile Repository SPT Satellite Position Table SSL Secure Sockets Layer SSM Source Specific Multicast ST Satellite Terminal SVN Satellite Virtual Network SVN-ID Satellite Virtual Network IDentifier SVNO Satellite Virtual Network Operator SW Software SYN SYNchronisation Sync Synchronization TBTP2 Terminal Burst Time Plan 2nd generation TC Traffic Class/Classifier TC/PHB Traffic Class/Per-Hop Behaviour TCP Transmission Control Protocol TCPM TCP Maintenance and Minor Extensions TDF Traffic Detection Function TDM Time Division Multiplexing TDMA Time Division Multiple Access TFO TCP Fast Open TIM Terminal Information Message TIM-B Terminal Information Message – Broadcast TIM-U Terminal Information Message - Unicast TIMU Terminal Information Message Unicast TIMu/TIM-U Terminal Information Message – Unicast TLS Transport Layer Security TMN Telecommunications Management Network TTL Time To Live ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 17 TV Television Tx Transmit/Transmitter TXID Transmitter IDentifier UDP User Datagram Protocol UE User Equipment UMTS Universal Mobile Telecommunication System USM User-based Security Model VACM View-based Access Control Model VBDC Volume Based Dynamic Capacity VLAN Virtual Local Area Network VoIP Voice over Internet Protocol VPN Virtual Private Network VRF Virtual Routing Forwarding WAN Wide Area Network WCDMA Wideband Code Division Multiple Access WFQ Weighted Fair Queuing XML eXtended Marked-up Language |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 4 Reference system architecture | Refer to [i.1] for the description of a reference system architecture. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 5 IP routing with OSPF over the satellite interface | IP routing is a key feature to realize seamless integration with terrestrial access networks and to attain multi-vendor interoperability in interactive broadband satellite networks. This clause provides review and recommendations for IP routing support in satellite networks over the satellite interface with emphasis on the Open Shortest Path First (OSPF) protocol for IPv4 (also called OSPFv2) [i.5]. OSPF performance is analysed under the different modes specified by the OSPF specification, namely, broadcast and non-broadcast modes. Satellite adaptations are proposed to improve OSPF performance in interactive broadband satellite networks. OSPF for IPv6 (also referred to as OSPFv3) [i.7] is also analysed to identify the required adaptations to improve its performance over interactive broadband satellite networks. The basic IP routing function forwards unicast packets according to the Routing Information Base (RIB). Compiling the information to form a Forwarding Information Base (FIB) can optimize this forwarding. The FIB is usually derived from routing information disseminated in the IP control plane, and stored in the RIB. The RIB is populated either via static configuration or dynamic routing protocols. Dynamic routing protocols are divided into interior (intra-domain) and exterior (inter-domain) protocols. The most common interior routing protocols are the Routing Information Protocol (RIP) [i.6] and Open Shortest Path First (OSPF) [i.5] and [i.7]. The most common exterior routing protocol is the Border Gateway Protocol (BGP) [i.8]. Although all routers provide routing functions, there are significant differences in their feature sets, often determined by where they are placed within the network. For example, customer routers often have reduced feature sets and may use static routing while enterprise routers have expanded feature sets, usually use OSPF and may support VPNs, VLANs, access control, network management, VoIP services, firewall functions and L2 virtualization. Satellite access networks have typically used static routing, although dynamic routing has been used for backhaul and restoration services in provider networks. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 18 |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 5.1 IP routing in satellite networks | A satellite network may support standard IP routing protocols using a L2 interface directly connecting external routers. Alternatively, the routing functions – adapted for satellite networks – may be integrated in the RCST. Dynamic routing for satellite is defined as the case where a routing protocol is used to route to the networks that are connected via routers to the LAN interface of a RCST. In contrast, static routing uses configuration information in RCST to determine the routing. Dynamic routing is attractive when the routing information is imported from connected networks. Many network operators implement policies to control the imported routes to protect the routing information in the RIB, although this is not recommended for link-state protocols, such as OSPF. Instead, BGP is widely used for exterior Gateway routing (between domains). BGP use is often accompanied by significant operator policy/configuration, and a need for integration of other protocols (e.g. tunnel management). As such, it may be more appropriate to place BGP functionality in an externally attached dedicated BGP router, rather than within a RCST. Such a router could accept exported routes from the satellite domain. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 5.2 Packet forwarding in satellite networks | An RCST may forward packets in one of the following modes: Static IP routing mode: The RCST acts as a router within an IP network. Forwarding is performed using the RIB or the compiled FIB. Static IP routing does not require a routing protocol; the routing information is derived from configuration and statically loaded into the RIB. In a star network, the RIB/FIB at a RCST normally has a default route that points to the Gateway, and configured routes at the RCST and GW to the networks connected via the RCST and GW LAN interfaces. Dynamic IP routing mode: The RCST acts as a router within an IP network using a dynamic routing protocol to populate the RIB. Dynamic virtual IP routing mode: A router that supports VRF groups maintains a set of completely isolated routing entities, one for each supported VRF group. Dynamic routing protocols, such as OSPF use link-local multicast, however many currently deployed satellite networks do not support this. Although the satellite outbound may allow RCSTs to receive multicast, the Gateway often does not replicate inbound multicast to all RCSTs. Internet links that do not support multicast routing should use point-to-point mode between adjacent routers. This mode cannot take advantage of the improved transmission efficiency offered by multicast. Static IP routing is the standard mode recommended for stub networks: static routing information may be distributed in the satellite network by configuration. It may also be possible to export routes from the Gateway to the RCSTs using standard OSPF routing packets. Support for dynamic routing requires correct treatment of link-local multicast (and at minimum, support for reception of these packets from the RCST by the Gateway/NCC). |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 5.3 Satellite network routing topologies | Figure 5.1 shows two example network topologies for OSPF. • Stub Networks. In this topology, Figure 5.1a, the RCSTs are stub routers that correspond to the scenarios normally expected for Consumer, SOHO, Multi-dwelling and Backhauling profiles. This topology assumes that the RCST is aware of (or allocates) the address space for the network connected via the RCST LAN interface. In this case, the NCC/Gateway already knows the remote network topology. This is expected to be often the case for star networks using private addressing, or where the LAN interface connects a remote NAPT or provides a point-to-point link that carries a tunnel (as in many Backhauling scenarios). ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 19 • Routed Stub-Networks. In this topology (Figure 5.1b), an RCST may have independently addressed sub- networks connected to the RCST LAN interface. This topology is expected to be common for Government and Corporate/Institutional use. It assumes that the RCST connects independently managed networks, where the address space for the network connected via the RCST LAN interface is not under the control of the NCC. In this case, the remote network topology is not directly known by the NCC and routing information is required to indicate which network address is reachable via which RCST. Dynamic or static routing is used to inject the routing information into the RIB at the satellite NCC/Gateway. The remotely connected networks may set a default route delivering all non-local traffic over their RCST air interface. • Routed Dynamic Networks. This topology consists of sub-networks connected to the LAN interface, and is the same as previous with one exception (Figure 5.1b). It connects networks using a dynamic routing protocol within the attached network. This is needed for networks that employ dynamic routing (e.g. to realize alternate paths to the satellite network, or where the satellite/alternate path is used as a backup for restoration of service following a failure). The important difference here is that the RCST should fully participate in the routing protocol exchanges (i.e. routing updates export topology information about the satellite network and import information about the reachable networks via the LAN interface). The key difference of this topology is that policies are needed for route import/export and whether the RCST router functions as a routing border. Figure 5.1: Example network topologies : : GATE WAY RCS T RCST RCS T SAT : : R1 R1 SAT GATE WAY RCS T RCS T R1 R1 a) b) R2 R2 R2 R2 R2 R2 R2 R2 ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 20 The use of dynamic routing in stub routers for a satellite star network is not usually necessary, since each RCST may be preconfigured with a static default route to the Gateway and routes to each locally attached network. The Gateway/NCC RIB would be statically configured to support the address range delegated to each RCST. Dynamic routing is desirable in cases where the set of networks reachable via an RCST can change without reconfiguration of the NCC. This may be the case for example when alternate paths exist (e.g. the connected network is also reachable via a terrestrial (backup) link) or where the NCC does not control the addressing plan (e.g. the operator of the remote network can move assigned IP addresses between sites without reconfiguration of the satellite network). |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 5.4 Dynamic routing using OSPF in transparent star networks | This clause describes the use of OSPF for dynamic IP routing in a satellite star network and provides guidelines on its use. The core OSPF algorithms are election of the Designated Router (DR), flooding of routing information, and OSPF route calculation. OSPF packets are directly encapsulated in the IP protocol using protocol number 89, using a combination of unicast and link-local multicast. OSPF sends Hello packets periodically on each interface to discover, establish and maintain a router's neighbour relationships. This also facilitates OSPF router configuration by indicating any support for optional capabilities. Routers of different capabilities can be mixed within an OSPF routing domain using the features advertised in this field. Note that Hello message exchanges serve to verify router configuration (e.g. adjacency, addresses used and protocol options) in addition to discovering neighbours and electing a DR. This functionality of Hello message exchanges makes it unattractive totally suppressing their transmission in a satellite network. OSPF supports various modes of operation. It runs in point-to-point, broadcast and nonbroadcast modes (in addition, virtual links can also be configured). Over non-broadcast networks, it can operate in one of two modes: non-broadcast multi-access (NBMA) and point-to-multipoint. Figure 5.2 shows the OSPF architecture in star and mesh systems. The first diagram corresponds to a star network, in which the DR is located in the Hub/GW. The OSPF function should give support for all the existing SVNs. The second diagram illustrates the mesh case, where each SVN has an RCST acting as GW, holding the DR. More complex architectures may include, in mesh networks, a GW supporting several traffic SVNs, giving OSPF support to these SVNs, resembling the transparent case. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 21 Figure 5.2: OSPF architecture for star and mesh satellite systems |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 5.4.1 OSPF for IPv4 | OSPFv2 is standardized as [i.5] and its extensions. OSPFv2 uses unicast and link-local IPv4 multicast. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 5.4.2 OSPF for IPv6 | Dynamic Routing for IPv6 is supported using OSPFv3 [i.7]. This standard replaces its earlier version, and removes support for dynamic multicast routing, instead preferring PIM-SM. OSPFv3 relies on IPv6 support on the router interfaces, including support for IPv6 link-local multicast. OSPFv3 presents some changes compared to OSPFv2 [i.5], mainly to support the increased address size and routing prefixes, and to provide a larger Options field. The core OSPF algorithms (e.g. DR election, flooding, OSPF calculation) remain unchanged. OSPFv3 also supports multiple routing instances on a link. The changes do not significantly impact the performance over satellite networks, the more compact format used in OSPFv3 results in approximately the same overhead even when using the larger IPv6 addresses. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 22 |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 5.4.3 OSPF Designated Router | In OSPF broadcast mode, a Designated Router (DR) (and backup DR, BDR) is elected via the exchange of Hello packets. Once elected, the DR sends multicast Hello packets every HelloInterval to other routers in the same area, whereas the rest of the routers send unicast Hello packets only to the DR and BDR. In other words, DR and BDR should be adjacent to the rest of the routers of the area. If a router does not receive a Hello packet from a certain adjacency during a RouterDeadInterval, it will declare the router down. In the topologies shown in Figure 5.1, in a satellite star network, the DR functions will normally be located at the Gateway/NCC. The BDR role may not be needed (because a Gateway failure results in the network becoming unavailable), or the BDR role may be co-located at another router also at the Gateway/NCC. 5.4.4 OSPF – NBMA mode OSPF supports non-broadcast multiple-access networks (NBMA). In NBMA networks, the DR and BDR are statically configured, that is dynamic discovery of neighbours is not performed in NBMA mode. In a satellite environment, the Gateway would usually be configured as the DR. It should be emphasized that, in NBMA mode, OSPF messages are always unicast; this would include OSPF messages on the forward link, which is an inefficient use of forward link capacity. 5.4.5 OSPF – Asymmetric multicast support A RCST may support link-local multicast transmission towards the Gateway, in which it can send IP multicast packets to the Gateway via the return link. A RCST may send to the multicast groups AllSPFRouters (224.0.0.5 for IPv4 and FF02::5 for IPv6) and AllDRouters (224.0.0.6 for IPv4 and FF02::6 for IPv6) but, in many current star satellite systems, this does not result in the retransmission on the forward link of these multicast IP datagrams that have originated from a RCST. Therefore, even though each RCST may send an OSPF Hello packet to the IPv4 group 224.0.0.5 (FF02::5 for IPv6), other RCSTs may not receive these packets. This means that RCSTs may be unaware of other RCSTs in the same OSPF area. In summary, it is recommended that each RCST have the capability to transmit link-local multicast packets to the Gateway, even if the RCST is not updated to support other multicast functions. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 5.4.6 OSPF RCST steps | Figure 5.3 shows the steps followed by an RCST that implements dynamic routing. The initial OSPF configuration (DR address) is obtained in the logon response message. The Hello protocol allows the creation of adjacencies and OSPF options configuration. During the OSPF flooding process, the RCST synchronizes with the SVN routing information and updates its RIB. Upon a LAN update (new public prefixes reachable through this RCST), the RCST propagates the new routing information towards the DR. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 23 Figure 5.3: RCST OSPF steps |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 5.4.7 Optimization of adjacency detection for satellites | In many deployed networks, the default OSPF configuration does not offer sufficiently fast detection of loss of connectivity to an adjacent OSPF router. One method to increase detection of this failure would be to send Hello packets more frequently and reduce the corresponding timers. However, this increases the overhead, and is undesirable for wireless/satellite links. Another method to quickly detect loss of connectivity to an adjacent router is to run a lightweight UDP protocol, known as Bi-Directional Forwarding Detection (BFD), specified in [i.9] and [i.10]. BFD can enhance detection of failures of an adjacency by providing a signal to the routing engine following a loss of a link. The BFD exchanges, while small, may also be undesirable for a satellite system, since these (like Hello packets) are sent irrespective of the traffic on a return link. In a satellite context, a similar gain to the use of BFD may be achieved through the use of a lower layer signalling mechanism that detects loss of the channel. This solution may save overhead by avoiding frequent packet exchanges. One optimization could be to eliminate use of Hello packets to increase performance. Loss of router adjancency over a satellite link could be quickly detected at the LL layer (below the HLS) and indicated to the RCST router without the need of IP-level Hello exchanges. Furthermore, a periodic exchange of Hello packets would also consume capacity on the return link, even for a RCST that carries little or no other return link traffic. However, it should be noted that Hello packets are not only used to verify adjacency (lack of which can be detected below IP), but also to notify the configurations supported through their Options field. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 24 The recommended solution is to reduce the periodicity of broadcast Hello packets from the Gateway, and to suppress Hello packets from RCSTs; except, when an RCST starts or restarts the OSPF routing process. This recommendation is motivated by a desire to reduce the capacity consumed by dynamic routing traffic. The cost of transmission on the forward link is much lower than for transmission on the return link from an RCST, which (in any case) could be idle. In addition, a single copy of the Hello packet is multicast to all RCSTs within a satellite virtual network. The periodicity may be reduced, since RCSTs do not need to use this to elect a DR (this is statically configured to be the Gateway), and an RCST may cache the options and the DR address from previous messages. The advantage of this approach is that it preserves the normal characteristics of the OSPF protocol – that is it provides a mechanism at the IP level to detect and confirm the adjacency, options in use, etc. Such confirmation provides logging information to an operator that ensures that incremental deployment of updates and the validity of any changes in configuration are noted at the IP network level (rather than being solely reliant on correct configuration of lower layers). The router memory requirement seems to be acceptable for the current designs of RCSTs. The satellite capacity consumed will depend on the topology and OSPF mode. Figure 5.4: OSPF packet exchanges at startup R1 (DR) R2 Hello (DR=0, None seen, Options) Hello (DR=R1,R2,R4 seen, Options) DD(seq= z, M) DD(seq= z, S) DD(seq= z+1, M) DD(seq= z+1, S) Hello (DR=R1,R4 seen, Options) Hello (DR=R1, R3,R4 seen, Options) Link State Request Link State Update Hello (DR=R1,R4 seen, Options) Hello (DR=R1, R2,R4 seen, Options) Link State Acknowledgement Hello (DR=R1,R4 seen, Options) Hello (DR=R1, R2,R4 seen, Options) 10sec 10sec 10sec GATEWAY R2 Hello (DR=GATEWAY, None seen, Options) Hello (DR=GATEWAY, R2,R4 seen, Options) DD(seq= z, M) DD(seq= z, S) DD(seq= z+1, M) DD(seq= z+1, S) Link State Request Link State Update Link State Acknowledgement Hello (DR=GATEWAY,R4 seen, Options) Hello (DR=GATEWAY, R2,R4 seen, Options) Configurable e.g., 1 minute, 1 hour. a) b) ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 25 Figure 5.4a) shows the default OSPF packet exchanges between a DR and a router, R2, in broadcast mode. Figure 5.4b) presents the packet exchanges for the satellite-adapted version. The traffic Gateway operates as the DR, and the Hello packets interval is updated to advertise the configuration and options supported. The election of DR in (b) will be suppressed since the Gateway Router will take this role. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 5.5 Dynamic routing for mesh satellite networking | In a mesh network, an RCST may be capable of direct communication with another RCST, without requiring the use of a Gateway to regenerate the signal. Although mesh communication enables traffic to be directly sent from one RCST to another, the underlying communication may be still under the control of a single NCC. Mesh networks may be divided into several categories: • Large Enterprise networks, e.g. with 200-2000 RCSTs controlled by a NCC. Such networks could be used for LAN interconnection, with the ability for any RCST to directly reach any peer RCST. The destination RCST could be identified by a layer 2 label. If this network operates at layer 2, then a next-hop resolution method is required to determine the value of the label to be used to reach a destination RCST. Routing is required to identify which RCST is to be used to reach a destination address assigned to a remote network. • Hybrid star/mesh, where traffic is routed via either star or mesh connectivity. An RCST could direct traffic via either a traffic Gateway or a mesh connection, depending on the intended destination. QoS-based routing is also possible. Example applications of using QoS to make routing decisions include routing VoIP over mesh (to minimize delay), data over a star connection (to minimize cost). Routing is required to identify which RCST is to be used to reach a destination address assigned to a remote network, the DR is assumed to be at the NCC/Gateway. • Small hub-less mesh. In this network, one RCST may assume the role of an OSPF DR (as in a star network, there may be a natural choice of RCST, if one RCST acts as a trafficGateway for the user traffic). This style of network may be well suited to applications such as SCADA networks. From a network perspective, this topology resembles that of a regenerative satellite network, where connectivity between RCSTs is possible without the role of a dedicated Gateway terminal. • Mesh networks over semi-transparent satellite where a satellite supports multiple types of transponder: one optimized for star networks and one optimized for mesh use. From a network-layer perspective, this resembles the hybrid star/mesh case. In summary, a dynamic routing mechanism is needed for mesh networks to direct traffic over the mesh connection, since the routes available depend upon the mesh capabilities. This mechanism is also needed to integrate mesh networks with terrestrial networks using OSPFv3. A possible solution may be to use standard IP-based methods for routing (as described in the present document) and to synchronize the detection of router adjacency with the lower layer functions that establish and release a physical layer stream. Two design options are presented to realize such a system: 1) The NCC is responsible for setup and clear down of mesh connectivity, and one approach could be for the co- located DR router to make all mesh routing decisions following establishment of a lower-layer mesh connectivity by the NCC. The establishment of a mesh connection would therefore trigger an OSPF routing update. 2) Another option is to establish the mesh connectivity at the physical and link layers via the NCC, and then for the RCST to exchange Hello packets over the established link, resulting a routing update from the RCSTs to the DR, and a corresponding routing update to all RCSTs. This interaction follows normal OSPF behaviour. The exchange of the Hello packets also establishes that the link is operational and permits the exchange of configuration data via the Hello packet options. Reachability can be validated by cross-layer mechanisms, eliminating the need for periodic Hello packets over an established mesh connection. Method-1 eliminates some RCST signalling, although it is less robust than method-2, since it does not validate the IP path. Method-1 also requires that any RCST policy (e.g. which addresses/traffic classes are to be routed via the mesh connection) needs to be configured and maintained at the NCC, rather than allowing the possibility that this could be a locally-configured RCST policy. Dynamic Connectivity Protocol (DCP) explained in [i.1] is recommended for setup and release of lower-layer mesh connections. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 26 Figure 5.5 shows an SVN routing update triggered by a routing change in the LAN connected to an RCST. This scenario requires multicast transmission of LSA updates from at least the RCST with the role of DR in the SVN. An RCST logon event should result in sending a Hello message and may also trigger the sending of routing updates by the DR. Figure 5.5: OSPF in mesh network The corporate scenario is characterized by many terminals and a medium-to-big GW including the DR. The RCST-GW LAN interface address is configured as the default IP next hop in all RCSTs of the SVN. This scenario may support multiple SVN groups by the GW. In case a dedicated RCST-GW is not used per SVN, the common RCST-GW should support OSPF independently on each VRF group. When the processing capabilities of the RCST-GW are not enough to host the OSPF router, OSPF forwarding capability by the RCST-GW towards the DR may be requested, unless the GW includes several RCST-GWs and the satellite AS is divided into smaller areas. This means forwarding OSPF packets to/from the satellite interface without decrementing the TTL field in the IPv4 header. In this scenario, the GW OSPF router is integrated with terrestrial networks, being the satellite network Area Border Router (ABR) and using BGP protocol. In a mesh network, it may be possible having a secondary (backup) DR, corresponding to a second RCST-GW, in an SVN. The OSPF routers may be or not co-located. In the example of having a GW including two RCST-GWs in active/stand-by configuration, when the RCST-GW1 fails, OSPF should restore mesh links between the RCSTs in the SVN and the RCST-GW2, which is the backup DR. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 5.5.1 OSPF and DCP | In mesh systems, OSPF packets are propagated over the mesh links established via DCP. For systems where the OSPF function is activated on the satellite interface, DCP only provides L2 address resolution function, and the IP routing information needed to construct the request messages should be provided by the source RCST, using the information in its RIB (dynamically updated by OSPF). A new DCP request will be issued by the RCST when there is no other mesh link opened directed to the same destination IP, in the same SVN, and using the same HL service. In other case, IP packets are forwarded to one of the opened links. Figure 5.6 shows a successful DCP exchange. An IP packet reaches the LAN interface of RCST1. Thanks to its RIB, RCST1 knows that the next hop IP address to reach the packet destination is the RCST IP router address and includes it in the DCP request message. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 27 If the source RCST supports multiple SVNs, the SVN mask included in the request should be used by the NCC to guess infer the SVN in which the mesh link is to be established. With this information and the destination IP address (or the next hop IP address), the NCC is able to locate the MAC24 address of the peer RCST, which will be included in the DCP response message. If the next hop IP address of an outgoing packet is not found in the AR database, a DCP Link establishment request is triggered by the RCST to find the L2 address of the next hop. In case that the system does not support the dynamic routing function (e.g. OSPF), the DCP protocol can assist the RCST with IP routing information. The NCC allows establishment of DCP Links only between RCSTs belonging to the same SVN or located in a common VRF domain, otherwise rejecting the Link requests. The RCST may indicate in the request message the next hop IP address (Next hop address field in the Triggering datagram identifier IE) according to its RIB. When this field has been filled by the RCST and the NCC cannot identify the destination RCST from the triggering packet destination address, then the NCC should use the address of the next hop field to obtain the MAC24 and the FPDU identifiers corresponding to the peer RCST. The transparent mesh RCST obtains the bursts to be decoded from its Active Links Table. The information about the other RCST Assignment_ID is obtained from the DCP messages sent by the NCC. It is assumed that, in a mono-beam transparent mesh system, all the RCSTs decode the same TBTP2 and therefore can extract information about the timeslots used by the other peer of the mesh link. Figure 5.6: Example of DCP exchange for bidirectional mesh link (addressing parameters) Figure 5.7 shows the internal procedures of an RCST supporting dynamic connectivity. The RCST should issue DCP requests when the next IP hop (corresponding to IP packet destination) is not found in the DCP active Links table. When found, the associated SVN number should match as well. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 28 Figure 5.7: IP routing and DCP |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 5.6 Recommendations for satellite routing support | This clause describes dynamic IP routing using OSPF in a satellite star network. OSPFv2 and OSPFv3 are used for IPv4 and IPv6-based networks, respectively. An implementation should adopt the standard mechanisms specified by the IETF, together with the updates recommended in this clause. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 29 Two topologies can be considered for transparent star satellite networks where the Gateway takes the role of the DR: 1) Use of OSPF may not be desirable for large satellite networks where all the RCSTs form one routing area. The advantages of dynamic routing are minimal where the addressing plan at the sites connected via the RCST LAN interface are expected to be under the control of the NCC. This is normally expected to be the case for Consumer, SOHO, Multi-dwelling and Backhauling scenarios. Therefore, static routing may be preferred in these cases. An RCST may be preconfigured with a static default route to the Gateway and routes to its locally attached networks. The routing tables at the traffic Gateway could also be statically configured to support the address range delegated to each RCST. Configuration of static routes could be made through standard IP management (e.g. SNMP, netconf, CLI), or could be considered as a part of the lower layer configuration. In IP networks, the former is usually preferred. 2) Dynamic routing using OSPF may be desirable for scenarios where the satellite network feeds one or more networks (e.g. using public address space) where the addressing plan is not under the control of the NCC. It is also desirable for networks that employ dynamic routing (e.g. to realize alternate paths to the satellite network, or where the satellite/alternate path is used as a backup for restoration of service following a failure). This is expected to be common for Government and Corporate/Institutional use. It is recommended that a star satellite network uses the OSPF Broadcast mode to take advantage of the lower transmission cost of sending multicast packets from the DR during the flooding process. This requires that link-local multicast packets originating from an RCST are sent on the inbound (return) link to the Gateway/NCC where the OSPF process executes. The remainder of this clause makes recommendations for optimizing performance in a satellite network. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 5.6.1 Recommendations for transmission of Hello packets | This clause updates the Hello processing described in [i.5] and [i.7]. This update applies only to a satellite interface. The reception of Hello packets by an RCST should be used to verify the correctness of the present configuration. It is desirable that IP functions are verified at the IP layer, rather than being entirely reliant on the correct configuration of end-points and lower layers. This promotes the Internet concept of "fate sharing", whereby a network path can be used if it is known to be functional, rather than relying on correct pre-configuration of protocols at lower layers. Hello packets should be sent by the Gateway using multicast to reach all RCSTs within a satellite virtual network. The periodicity of these messages may be reduced depending on operator needs, but should not be reduced to less often than every 30 minutes (a value chosen as a trade-off between ability to detect misconfiguration and overhead). Hello packets received by an RCST should be used to verify the correctness of the present OSPF routing configuration, including checking the DR IP address and options. It is desirable that IP configuration values are verified at the IP layer, rather than being reliant on correct configuration of end-points and lower layers. An RCST should cache the last received Hello packet. This optimization may enable it to quickly restart the OSPF process with a relevant configuration. If used, this cache should be cleared when the NCC restarts the satellite network following a configuration change that affects IP routing (e.g. change of addressing plan or change in Gateway router configuration). It is recommended that an RCST sends a single Hello packet to the DR during the "set-up" phase of an RCST. This packet indicates the router's capabilities through the Options field. Hello packets may also be exchanged at periodic intervals to verify the state has not been changed, but this interval need not be small (as in reachability detection), and a much longer value may be configured (e.g. at intervals of minutes/hours). Note that sending this packet once does not constitute a major overhead, but confirms reachability. A drawback is that this packet could be lost in the network i.e. the configuration soft-state is not refreshed by the protocol. Network operators need to be aware of this possibility when managing their networks. This does not impact dynamic routing, provided that the configuration of RCSTs and Gateway are consistent. The RCST should resend this packet if the IP routing configuration of an RCST changes or the OSPF process is restarted. An RCST that detects a loss of the forward link or a state transition at the lower layer that prevents IP transmission on the return link should update its OSPF adjacency as if the router at the RCST had failed to receive a Hello packet. This failure detection is similar to the use of BFD in [i.71]. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 30 |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 5.6.2 Recommendations for routing topology update | It is recommended that standard OSPF methods are used to propagate the routing information, and RCSTs are enabled to send these updates using link-local IP multicast to the Gateway router. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 5.6.3 Recommendations for defining OSPF Areas | The signalling cost of using dynamic routing is highly dependent on the topology of the network. Scaling, i.e. the number of routers using a single DR needs to be considered to minimize routing traffic. Judicious configuration of border routers (ABRs) to divide routing areas is recommended at an RCST, where the RCST connects more than a few OSPF routers via the LAN interface. This may be especially useful when the routing information may be summarized or when there are frequent routing updates within the network fed by an RCST router LAN interface. The principles for configuring ABRs are not specific to satellite and advice can be obtained from usage in other networks. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6 IP multicasting | This clause elaborates on the MMT2 supported method for mapping multicast to L2 as being the most versatile method of the two alternatives specified by [i.3]. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.1 Mapping IP multicast to L2 | |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.1.1 Multicast over Ethernet | Modern Ethernet controllers filter multicast frames out of received frames to reduce the host CPU load. This is achieved by deriving a L2 multicast group destination address for each IP multicast group that needs to be received/forwarded. The set of active L2 addresses is stored in a table maintained by the host software. This table is used to decide whether a multicast frame received on an interface is forwarded to L3 or discarded. The L3 address is also checked against the set of groups to be received before the packet is forwarded to higher layers for further processing. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.1.2 Mapping of IPv4 addresses | The mapping between IP and L2 multicast group addresses at the LAN interface is usually provided by a standard method [i.11] that derives a L2 address for each IPv4 multicast group destination address. This method is commonly used for all multicast networks, and it is also the method specified for GSE satellite systems that use the 6-byte address format. [i.3] specifies two alternatives for mapping multicast to MAC24, one operating without support of MMT2 and one using MMT2. The NCC determines which method that applies, and provides the necessary configuration as part of the LL service configuration. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.1.3 Mappings for IPv6 address | IPv6 includes multicast as a standard function, and many core IPv6 protocols rely upon multicast support. The same IPv4 multicast requirements exist for IPv6. However, there are some additional considerations: 1) IPv6 multicast redefines the way scoped addresses are specified, this places additional constraints on filtering addresses when forwarding between different networks (or virtual networks). 2) Some protocols (e.g. neighbour discovery) generate link-local IPv6 multicast addresses, which means that many LANs carry a large range of IPv6 multicast groups, none of which is intended to be forwarded by a router. 3) The IPv6 address range is larger than that of IPv4, which can result in more overlap of addresses (i.e. two addresses in different address blocks can map to the same L2 address). For dual-stack deployments with significant levels of both multicast IPv4 and IPv6 traffic, it is recommended that separate L2 address spaces are used for the two services to avoid address overlap. In cases with lower levels of multicast traffic, or where the addressing plan is under the control of the operator (who could assign addresses to avoid overlap), the two protocols (IPv6 and IPv4) may share the same L2 address space. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 31 [i.3] specifies two alternatives for mapping multicast to MAC24, one operating without support of MMT2 and another using MMT2. The NCC determines which method that applies, and provides the necessary configuration as part of the LL service configuration. 6.2 Operator-controlled mapping of Layer-2 multicast group addresses The mapping between L3 and L2 addresses specified in [i.11] may result in two L3 groups may map to the same L2 address, which is usually referred to "address overlap". This is not normally a concern for LANs thanks to the additional IP multicast group address filtering at the IP layer. It can be a significant issue when two completely different services map to the same address, since the forwarding is usually controlled per L2 address. This can, for instance, occur when the network link carries traffic from more than one service operator (e.g. in a multicast Internet exchange point). One way to avoid the issue of address overlap is by careful L3 address assignment. This is recommended in static configurations although hard to manage with dynamic multicast. It is common for applications to choose well-known IP multicast addresses, and this would result in unexpected behaviour when more than one virtual addressing space is used. Separation of different multicast address scopings is essential for proper multicast operation [i.12]. DVB-RCS2 may provide support for multi-operator use of multicast. This addresses a need to support Internet multicast access or/and when multiple virtual networks are supported over the satellite, by allowing a satellite virtual network operator to control the mapping to L2 address. The mappings are configured in the Feeder (along with any required QoS requirements). This device is managed by an SNO and also coordinates the mapping for unicast network-layer packets for each SVNO. The mappings configured at the Feeder are also announced by using a control table called the DVB Multicast-Mapping Table (MMT2). This is organized into a set of sections directed to each SVN that supports multicast. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.3 IP multicast forwarding over satellite and LAN interfaces | This clause explains the use of IP multicast control techniques and the use of the Internet Group Management Protocol (IGMP) [i.13], [i.14] and [i.15] to deliver multicast content to an RCST. Equivalent behaviour is also expected for IPv6 using the Multicast Listener Discovery Protocol (MLD) [i.15], [i.16] running over ICMPv6. Support for IGMP over the LAN interface can be classified as either passive or active. For DVB-RCS2, the active mode refers to use of an IGMP or MLD intercepting proxy agent operating over the LAN interface at the RCST. Multicast receiver hosts do not participate in routing decisions, and instead use multicast control protocols to signal the set of groups that they wish to receive. IGMP is used for IPv4 and MLD is used for IPv6. Upon reception of these control messages, a multicast router triggers appropriate routing messages (e.g. PIM-SM messages [i.17]) to control forwarding from any routed upstream network node that supports IP multicast. Managed Ethernet switches typically implement an IGMP Proxy [i.18], in which an agent intercepts membership reports from hosts and uses this information to determine over which LAN interfaces to forward IP multicast packets. This is similar to the way a multicast-enabled IP router processes these reports. An RCST may be configured to function in either a passive or an active multicast control mode in regards to how they forward multicast traffic on its LAN interface. In passive mode, no IGMP messages are processed by the RCST and no multicast membership reports are sent on the satellite interface, whereas in active mode IGMP messages are terminated at an IGMP proxy agent in the RCST and may then be sent over the satellite (when dynamic multicast forwarding is used). In a satellite network, an RCST implementing active mode over the LAN interface also uses a proxy agent that participates in the multicast control protocols, to populate a local data structure identifying the set of presently active IP multicast groups (the Multicast FIB, MFIB). In contrast, a RCST using passive mode over the LAN interface provides a configuration interface to insert entries in the MFIB. When using passive mode, an RCST forwards traffic to the LAN interface independent of whether there is an active receiver on a connected host. This mode resembles broadcast, in that the Service Provider determines the forwarding. In many cases it is not necessary that all multicast groups are forwarded to the LAN interface by all RCSTs, and in passive mode this is controlled by RCST configuration. Downstream devices connected to the LAN Interface of an RCST such as managed Ethernet switches, may control the propagation of specific groups using standard methods such IGMP snooping [i.19] or IGMP proxy [i.18]. IGMP Proxy is generally preferred. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 32 The procedures for multicast traffic sources attached to RCSTs and bidirectional multicast are not defined in DVB- RCS2. The current document also does not specify multicast router interactions with an RCST. Neither does it describe the support of multicast routing over the satellite interface. The preceding text focuses on multicast forward control at the RCST over the LAN interface. The remaining text delves into the forwarding multicast traffic that is received on the satellite interface. The NCC is responsible for all multicast transmission on the forward link. Prior to multicast data transmission on the forward link, the NCC should first configure the Feeder and the Gateway Router with entries for those multicast streams that are to be forwarded. Once configured, the Gateway Router joins the requested group at its upstream interface. The feeder encapsulates the multicast traffic and forwards it over the satellite air interface. Static and dynamic forwarding are distinguished in this case: In static forwarding, this process is completed before the RCST needs to receive the multicast group. In dynamic forwarding, this process is completed the first time an RCST requests to join a specific group. An RCST that wishes to receive multicast traffic with a specified IP address on the forward link should first construct a layer 2 filter containing the GSE labels with which the multicast traffic is sent. This table may be directly mapped using the information in the MMT2 to identify the GSE address used to carry a multicast flow. Once configured, the filters forward all traffic with the label to the IP layer where the RCST filters the traffic based on the IP network layer address using the information in the MFIB. Static forwarding on the satellite interface may be extended by enabling a proxy agent at the RCST to provide active mode at the LAN interface. Static forwarding on the satellite interface and active mode on the LAN interface is intended to be the default for DVB-RCS2 RCSTs. The proxy agent intercepts a group management protocol (e.g. IGMP, MLD), by intercepting packets received on the LAN interface to build a local forwarding table (e.g. held in the Multicast Forwarding Information Base, MFIB). Multicast traffic received from the forward link is only forwarded to the LAN interface when there are active receivers connected via the LAN interface that need to receive the specific IP group. This prevents the LAN from having to carry traffic for services for which there are no receivers, an attractive optimization provided in most multicast networks. Static forwarding over the satellite has advantages in terms of simplicity of design of the Gateway and control of the QoS offered to each multicast group. However, the approach relies on the operator determining what content is to be received at any time. While this is appropriate for pre-scheduled transmissions (such as file updates, IPTV broadcast, etc.), it is not appropriate for applications that are user-driven (such as video-on-demand, multi-party conference, collaborative working applications and service discovery). User-driven applications often cannot predetermine the set of multicast groups that will be used, and it is often not feasible to forward all multicast traffic over a satellite, irrespective of whether there are any active receivers for the given groups. Dynamic forwarding on the satellite interface is required in these cases to control the set of groups that are forwarded from the Gateway to the receivers. When dynamic forwarding is used, each RCST determines – and, indicates to the NCC – the multicast group traffic that it wishes to be forwarded to it. This information is collected by the Proxy by reception of IGMP/MLD Membership Reports on the LAN interface. The RCST indicates the need to forward traffic over the satellite by sending a group membership message (join) upstream to the Gateway. The protocol used for this control message could be an extension of the protocol used on the LAN interface – i.e. a RCST could proxy IGMP, MLD or PIM, or it could summarize the group membership state into a satellite-specific protocol. The flow of control information from the set of RCSTs with active receivers allows the Gateway to identify the set of groups that need to be forwarded over the satellite interface, and provides an indication to the NCC for control of the multicast service. This means that satellite capacity is not used for traffic for which there are no active receivers and, that a Gateway router could itself generate an upstream PIM-SM Join message for a requested group to dynamically request the content from a source connected via an upstream-multicast network. Dynamic forwarding increases the complexity of the multicast service as it necessitates more control functions to realize an effective operational service. For example, the set of groups requested by a RCST needs to be controlled by the NCC to prevent an RCST requesting unauthorized (i.e. outside the SLA for the RCST service) or illegal (i.e. for address ranges that cannot be used) groups. Additional multicast control functions may need to be related to Authentication, Authorization and Accounting, AAA, functions at the Gateway, to implement subscriber control and enable accounting for billing. The control functions need to be virtualized, if the same multicast traffic is to be forwarded to multiple SVNs. The mappings of content to lower layer addresses may also need to consider address translation when source-specific groups are used with private addresses. An SLA is also needed for each multicast service, to determine the QoS attributes to be used by the Feeder for each group or for all multicast traffic (e.g. peak rate, binding to L2 address and Lower Layer Service on the forward link). ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 33 Table 6.1 summarizes distinct forwarding modes discussed so far. Following clauses present more details and examples on these forwarding modes. Table 6.2 summarizes the advantages and disadvantages of dynamic multicast. Table 6.1: Multicast modes at satellite and LAN interfaces Mode Satellite Forwarding RCST Receiver LAN Forwarding Static, passive LAN forwarding Determined by NCC configuration. Independent of active RCST receivers. Static by configuration Independent of the set of active receivers downstream to the RCST. Static, Active LAN forwarding Determined by NCC configuration. Independent of active receivers. Forwarding decision made using IGMP/MLD proxy Only when there are active receivers downstream to the RCST. Dynamic (implies Active forwarding) Determined by Gateway informed by proxy at RCST, only for 1 or more active RCSTs. Forwarding decision made using IGMP/MLD proxy Only when there are active receivers downstream to the RCST. Table 6.2: Advantages and disadvantages of dynamic multicast Advantages Disadvantages The satellite capacity is only consumed for multicast traffic that is required at the receiver. Increased inbound control traffic when STs request to receive a multicast group, however the control traffic is usually much less than the total traffic. An RCST can receive arbitrary multicast flows (if permitted by the NCC), including traffic with an IP group destination address not known a priori. Increased complexity at the Gateway, where multicast control protocols (e.g. PIM) need to be deployed to communicate with upstream multicast networks. It does not require pre-configuration of the Feeder and Gateway router to support specific groups. The NCC may need to support dynamic construction of the MMT2 and reconfiguration of the Feeder. Provides the service operator with the ability to monitor/charge users for consumed content. Increased operational complexity in managing and supporting the service. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.3.1 Static forwarding with Passive mode on the LAN interface | In the passive multicast forwarding mode, the RCST does not participate in IGMP or MLD. Figure 6.1: IGMP passive mode for DVB-RCS with static forwarding Figure 6.1 shows the passive mode implementation of IGMP for the DVB-RCS2 network. The Gateway has no IGMPv2/v3 stack. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 34 Figure 6.2: Static multicast with passive forwarding using MMT2 Figure 6.2 shows a ladder diagram illustrating the delivery of multicast using static forwarding. The NCC configures the Feeder with the mappings for the MAC24 to be used for each active IP multicast address. In static multicast, the NCC will have previously generated MMT2 entries required and will have configured the Feeder and Gateway Router to forward the multicast stream over the forward link. The RCST will forward the multicast streams on its LAN interface if it is configured to do so, i.e. if the IP multicast address is in the MFIB. Otherwise, it will silently drop the multicast streams being received. An RCST may inspect the content of the MMT2 to identify all active GSE multicast mappings for the SVN to which it belongs. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.3.2 Static forwarding with Active mode on the LAN interface | Figure 6.3 shows the active mode for IGMP in a DVB-RCS2 network. In active mode, a proxy agent [i.18] is implemented at the RCST. The proxy implements timer values and forwarding rules associated with this active mode. In active forwarding mode, an IPv4 RCST performs the IGMP router function on its LAN interfaces and the host function of IGMP on its return satellite interface. An RCST should not perform the router portion of IGMP on its return satellite link. However, in the IGMP active mode, the Gateway can be queried on both of its interfaces (forward satellite link as well as the upstream core-network interface). Figure 6.3: IGMP Active-Mode for DVB-RCS2 with Static Forwarding ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 35 Figure 6.4 provides a ladder diagram showing the delivery of multicast in static configuration mode with active forwarding at the RCST. The only difference in this scenario compared to the previous, is the control of the static multicast streams at the RCST LAN interface. The RCST Agent collects IGMP/MLD Membership Reports received at the LAN Interface to populate the MFIB with the IP group multicast addresses of the traffic to be received by hosts on the LAN. It uses the information in the MFIB to forward the static multicast streams. This method is the proposed default case for the DVB-RCS2 system. Figure 6.4: Static multicast with active forwarding using MMT2 |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.3.3 Dynamic forwarding with Active mode | When a host sends a multicast membership report to the LAN Interface, the proxy agent at the RCST will forward the request upstream to the Gateway. The Gateway may forward this to the NCC to check the authentication and record the activity. In dynamic forwarding, the requests from the RCST trigger the Gateway Router to join a specific group on its upstream interface. When necessary, the Feeder is also (re) configured to forward the group, and the NCC may update the MMT2 to reflect any changes made to the Feeder configuration. Figure 6.5 depicts a ladder diagram for the case of dynamic multicast when the content is already being sent over the satellite air interface (e.g. when a different RCST has requested reception of the same IP multicast group). In this case the NCC has already configured the Feeder and the Gateway Router with entries for the required multicast stream to be forwarded. Figure 6.6 shows the ladder diagram for dynamic request of new multicast content. In this scenario, the content is not sent via the satellite until requested. A host sends an IGMP/MLD membership report to the RCST, which then forwards the request over the satellite link to the Gateway. Since the Gateway does not have an entry for this particular group, it requests authentication from the NCC. When it receives authentication, the Gateway Router send a multicast join request upstream to request the content. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 36 The NCC also performs any required update to the configuration of the Feeder and MMT2. If the update modified the MMT2 address mappings, then an updated MMT2 is required to allow the RCST to receive the multicast content. Once multicast traffic arrives at the Gateway Router, it forwards the content to the Feeder using the Forward Link. This scenario is different to that of the previous case because it may trigger configuration of the Feeder and the authentication of the request by the NCC, before a user receives the multicast stream. Pre-configuration of the Feeder can simplify the control interaction. For example, the Feeder could be allowed to forward a block of multicast addresses from the Gateway Router to a specific SVN and could advertise this binding in the MMT2 before it receives any request. This removes the need to reconfigure the feeder each time a request is received for a group address that was mapped. Figure 6.5: Dynamic multicast request for existing IPv4 multicast content ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 37 Figure 6.6: Dynamic request for new IPv4 multicast content Figure 6.7 shows the case where the authentication step results in rejection of a request by the NCC. When the RCST receives a multicast membership report it forwards this to the Gateway, which will then seek authentication from the NCC. If the NCC rejects this request, it will update the Gateway (and the RCST) to ignore all Join requests for the particular multicast group for a specified period. If the RCST receives additional membership reports, requesting to join the group during this interval, they will be silently dropped. It is important to note that the RCST should still forward the membership report for other groups during this interval. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 38 Figure 6.7: Dynamic request for IPv4 multicast content rejected by the NCC |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.3.4 IP multicast walkthrough in DVB-RCS2 | The following entities are required to support a DVB-RCS2 multicast service: • A multicast-enabled Gateway Router, that may use PIM-SM to request upstream content from the terrestrial network to which it is connected and forward this to the Feeder. • For the forward link user plane, the NCC should authorize use of forward link satellite capacity for a multicast service by the Feeder and coordinate the use of layer 2 addresses. • For the forward link control plane, the NCC may also need to generate a set of MMT2 tables for transmission by the Feeder indicating mappings for each SVN that it supports. • The Gateway receiver, in the case of dynamic multicast should process requests for content (join messages) from an RCST. • The Feeder should encapsulate and forward multicast flows on the forward link. The feeder also distributes the MMT2 control table to all RCSTs. • In the user-plane the RCST should enable multicast reception (filtering), forwarding and the processing of multicast address bindings, including parsing of the MMT2. • In the control plane, the RCST may also need to support an IGMP proxy function and use this to control forwarding and for the dynamic case, return control information to the NCC (the join message). ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 39 |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.4 Encapsulation of IP multicast packets | The functions required for multicast forwarding (see clause 6.3) over a DVB-RCS2 network can result in a range of system designs. • Static Multicast & Autonomously synthesised MAC24 address: IP traffic at the Feeder is mapped directly to a MAC24 based on the IP group destination address. The NCC enables forwarding of this group by the Feeder. An RCST that is set to receive an IP address (e.g. in the MFIB) maps the address to the corresponding 3B GSE label and then will unconditionally forward all traffic received with the configured IP address to the LAN interface. In static multicast all groups may be forwarded if this is configured. • Static Multicast & MMT2: The NCC associates a multicast MAC24 for each multicast group. This is used by the Feeder to set the 3B label in an encapsulated multicast packet. The NCC inserts an entry for the IP multicast group address for the corresponding MAC24 in the MMT2. This table is periodically sent by the Feeder to all RCSTs within an SVN. RCSTs are configured to receive IP groups by determining the IP address to be forwarded (e.g. in the MFIB), and then binding this to a MAC24 (e.g. using the MMT2). In the static case, this may default to all advertised multicast content, or could be restricted by local configuration. • Static Multicast, Active mode LAN interface: As above except the RCST implements a proxy agent. An RCST that receives signalling on its LAN interface adds the corresponding L3 group destination address to its local multicast forwarding state (e.g. in the MFIB). This requires the agent to respond to IGMP/MLD group membership on its LAN interface. Then, it determines the set of 3B MAC24 address required to receive the multicast streams, based on the MMT2. Packets received on one of these MAC24s are then filtered at the IP level based on the local multicast forwarding state, and all traffic that matches is forwarded over the LAN interface. • Dynamic Multicast: IP traffic at the Feeder is mapped directly to a MAC24 derived from the IP group destination address. The NCC will enable forwarding of this group at the Feeder. All packets of a flow not requested by any RCST will be discarded. The NCC will usually implement a policy to control whether a particular group is enabled and set corresponding QoS parameters for transmission. As in active forwarding, each RCST implements a proxy agent. This controls forwarding to the LAN interface, in addition, the RCST summarizes its local IP forwarding state (from the MFIB) to the NCC to allow it determine which group should be forwarded. An RCST that implements a Proxy should filter packets initially by MAC24 and finally by IP multicast address, ensuring that only the traffic belonging to the configured group(s) is forwarded (i.e. discard any unwanted traffic that maps to an overlapping MAC24). The present document does not include consideration of the design of the IGMP/MLD proxy or the PIM-SM router. Neither does it make recommendations on how these protocols should be configured/ adapted to the satellite case. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.4.1 Address mapping for IPv4/IPv6 addresses using the MMT2 | Multicast network addresses used in SVNs that belong to different VRF Groups may be identical but correspond to different multicast groups and need to be handled separately (i.e. avoiding address overlap). SVNOs may also need more control of the mapping used in their SVN, e.g. for traffic engineering or to minimize the cost of multicast filtering at RCSTs. An RCST can identify the MMT2 entries applicable to itself by monitoring the "svn_number" field in the received MMT2 (see [i.3]). The svn_number field is indicated to the RCST during logon. • An RCST will decode all entries in the MMT2 that match its pre-assigned "svn_number". The MMT2 comprises a set of entries corresponding to blocks of multicast addresses; the number of blocks is specified by the "mapping_sections" field. • Each block is specified for one GSE Type field. This value is used to differentiate address blocks defined for IPv4, from the ones for IPv6. The address size will be set accordingly. • For each block, the MMT2 specifies the start and end address. It may also specify values for specific multicast ranges as exceptions to a previously defined multicast address range. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 40 • For each block of address the RCST derives a base address value and length. The IP multicast address is mapped to a 3-Byte MAC24 address, by first identifying a MAC24 base address. The mcast_prefix_length defines the number of significant bits (starting from the most significant bit) that are used from this base address. The remaining bits of the 3-Byte MAC24 address will be mapped directly the low-order bits of the required IP multicast address. The simplest MMT2 contains one record per SVN that indicates one address block for the MAC24 range (i.e. mask size). In many cases, the mapping from network multicast address to MAC24 can be fairly static, with the mapping revised from time to time, when needing to add/remove a flow, or re-assign an existing flow to a new multicast address. In DVB-RCS2, the table is created at the NCC and used by the Feeder. Procedures may be defined to automate creation of the MMT2 entries, for instance, to create a mapping based on reception of a dynamic multicast join request from a SVNO (e.g. generated as a result of arriving multicast traffic or reception of IGMP/MLD messages). The MMT2 structure allows an RCST to map an IP group destination address to any SVN. In normal operation, an SVNO is expected to use a part of the address allocation for the SVN to support the multicast services. In this case the base address retrieved from the MMT2 for a multicast group will likely be a subset of the SVN address range used by the RCST for its unicast service. The MAC24_base address and mcast_prefix_length can be configured to suit the various multicast scope required by the satellite virtual network operators or the satellite network operator. Note that this scheme is not designed to allow a change of the mcast_prefix_length while RCSTs are logged onto the system. The recommended procedure is therefore to reset the mask length and, for users to then force RCSTs to logon to the updated system using the new mask length. This design decision is justified in that it is not expected that reconfiguration will be required. Figure 6.8: Example MAC24 address allocation to SVNs The MMT2 uses a 16-bit encoding to represent the SVN number used by a Satellite Virtual Network Operator. This identifies which systems read the MMT2. Note that the SVN-ID is encoded as a 16 bit number, corresponding to the two most significant bytes of the lowest used MAC24 in the allocated block. This is the form used in the MMT2 to advertise the multicast prefixes, and is used so that the network can accommodate multiple sizes of svn_mask. This allows an SVNO to subdivide a single Operator Virtual Network (OVN) allocation from an SNO to realize multiple SVNs within its assigned OVN. Hence, if an SVNO is allocated 0x0100000-0x01FFFF, the MMT2 entry may be 0x0100, but if the operator chooses to subdivide this OVN allocation into two, then he could separately generate a MMT2 for 0x01800 and 0x01000, since RCSTs always the know the SVN_mask length, they know whether the OVN has been subdivided in this way. SNOs will allocate SVNOs with "svn_numbers", which are then used by them to assign address ranges amongst their users (RCSTs). During startup RCSTs are informed of their "svn_number", SVN-MASK length and assigned a unicast MAC24 by the SVNO. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 41 Each SVNO can allocate parts of the allocated address as either unicast, multicast or reserved for future use. The allocation to be used by a particular SVN is notified in the Multicast Mapping Table 2 (MMT2). The MMT2 contains a list of mappings for multicast addresses for each "svn_number", i.e. for each SVN. An RCST receives the MMT2 using multicast. It examines "svn_number" of received messages and only accepts messages where the number matches a svn_number value assigned to one of the RCST interfaces. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.4.2 Mapping for IPv4/IPv6 addresses to the same MAC24 prefix | The MMT2 mapping should by default use different MAC24 prefixes for IPv4 and IPv6 traffic. This use resembles the use in Ethernet in GSE, where overlapping between address ranges do not exist, because IPv4 and IPv6 traffic is assigned to a different Organizationally Unique ID, OUI. In some cases, SVN address space can be conserved by mapping the two sets of IP addresses to the same MAC24 base address. However, this can also result in overlap between IPv4 and IPv6 multicast. This overlap could have unwanted side-effects. One scenario where this separation is desirable would be when a content provider maps content to both IPv4 and IPv6 and both use a format where the least significant part is the same. When an RCST dynamically registers for IPv4 membership join, it opens the L2 filter to receive the content. A subsequent request by an RCST that desires IPv6 content would then lead to this additional traffic being sent and would be passed by the existing L2 filter, requiring protocol filtering at the higher layer. While this filtering will only forward the requested IPv4 traffic, the IPv6 group still contributes to additional unnecessary processing cost at both L2 and L3. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.4.3 Aliasing for IPv4/IPv6 addresses using the MMT2 | An SVNO may decide through bilateral agreement with another SVNO or the SNO to use a MAC24 multicast prefix that lies outside the unicast address range that it uses. In this case, the base address retrieved from the MMT2 for a multicast group will belong to a different SVN address range. This effectively allows one operator to signal use of another block of addresses. This method could be used to group together multicast services for several SVNs and may eliminate the otherwise need to replicate common multicast streams for different services. Care needs to be exercised in using this method so that the addresses remain in scope (i.e. the aliased address has the same meaning in the SVN for which it is to be received). Hence, the MMT2 may be used to support a network group that is accessible from more than one SVN and is mapped to a common MAC24. The SNO/SVNO is responsible for such system-wide co-ordination of the use of MAC24 labels. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.4.4 Example address mappings using MMT2 | |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.4.4.1 Simple MAC24 mapping for multicast address blocks | Table 6.3: SVN 0x01 Example address block allocation SVN 0x01 Address Block (mcast_prefix_length = 9 bits) Allocation 0x010000 – 0x017FFF Unicast 0x018000 – 0x01FFFF Multicast The MMT2 defines a multicast SVN block using the "mcast_prefix_length", adding a single bit to the existing 8-bit (in this example) SVN-MASK (i.e. 9 bits). In Table 6.3, a SVNO that has been assigned 0x0100 and has divided its address block of 64K (approx.) address into two address blocks of 32K addresses for unicast and 32K for multicast. The SVNO therefore assigns a "mcast_prefix_length" of 9 bits. Table 6.3 shows the allocation of SVN 0x0100 with a different division of the address block. The Feeder will be configured with this allocation. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 42 Table 6.4: MMT2 example for SVN 0x0100 MMT2 Field Value Length svn_number 0x0100 16 bits pt_count 1 8 bits protocol_type 0x800 16 bits address_size 0x04 (4 Bytes) 8 bits mapping _section 1 8 bits inclusive_start 224.0.0.1 32 bits (4B) inclusive_end 239.255.255.255 32 bits (4B) exclusions 0 8 bits mac24_base 018000 24 bits mcast_prefix_length 01001 (9 bits) 5 bits The MMT2 shown in Table 6.4 is an example of the most basic scenario of multicast address mapping to a MAC24. In this example, the SVNO has mapped the complete multicast address range to a block of addresses within its address range. Table 6.4 illustrates the MMT2 used for SVN 0x0100. In this scenario, the Feeder generates the MMT2 during initial configuration and no further updates will be necessary to support requests for dynamic multicast streams because the entire multicast address block has been assigned. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.4.4.2 Dynamic MAC24 mapping for multicast address blocks | This example provides a use case for a dynamic multicast stream, where the Feeder will map the IP multicast address to an assigned MAC24 label in the range 0x02C000 – 0x02FFFF (Table 6.5). The value of mcast_prefix_length changes to 10 bits in this use case. The Feeder will transmit the MMT2 for the SVN (0x0200) to inform the RCST of mappings for the multicast stream. Table 6.6 shows the MMT2 for SVN 0x0200 in which the RCST is informed of the IP multicast mapping to a MAC24. Table 6.5: SVN 0x0200 Example address block allocation SVN 0x02 Address Block (mcast_prefix_length = 10 bits) Allocation 0x020000 – 0x023FFF Unicast 0x024000 – 0x027FFF Reserved for future use with Unicast or Dynamic Multicast 0x028000 – 0x02BFFF 0x02C000 – 0x02FFFF Multicast Table 6.6: MMT2 example for SVN 0x0200 MMT2 Field Value Length svn_number 0x0200 16 bits pt_count 1 8 bits protocol_type 0x800 16 bits address_size 0x04 (4 Bytes) 8 bits mapping _section 1 8 bits inclusive_start 224.0.0.1 32 bits (4B) inclusive_end 239.255.255.255 32 bits (4B) exclusions 0 8 bits mac24_base 02C000 24 bits mcast_prefix_length 01010 (10 bits) 5 bits ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 43 |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.4.4.3 MAC24 mapping using the "exclusions" field | The MMT2 syntax permits flexibility for mapping an IP multicast address to MAC24 labels. The "exclusion" field allows a SVNO to specify a different MAC24 mapping behavior for different IP multicast address ranges and assign these to MAC24 labels, i.e. dynamic allocation. Table 6.7 shows an MMT2 for SVN 0x0300, which has the same format of address block allocation as SVN 0x0200 (shown in Table 6.5). In this SVN (0x0300), the SVNO has used the "exclusion" field to exclude the SSM address range (as shown in Table 6.7). A use case may be to support dynamic multicast, where the MAC24 is assigned when multicast forwarding is setup. In this example, the Feeder dynamically assigns one SSM address with an MAC24 address from the unallocated range (shown in Table 6.7) with the use of a second "inclusion_start" and "inclusion_stop" section within the MMT2. A second "rcs_mac_base" field along with the "mcast_prefix_length" is used to assign an MAC24 labels from an address block different to that allocated in the first section. Table 6.7: MMT2 example for SVN 0x0300 MMT2 Field Value Length svn_number 0x0300 16 bits pt_count 1 8 bits protocol_type 0x800 16 bits address_size 0x04 (4 Bytes) 8 bits mapping _section 2 8 bits inclusive_start 224.0.0.1 32 bits (4B) inclusive_end 239.255.255.0 32 bits (4B) exclusions 1 8 bits exclusion_start 232.0.0.0 32 bits (4B) exclusion_stop 232.255.255.255 32 bits (4B) mac24_base 03C0000 24 bits mcast_prefix_length 01010 (10 bits) 5 bits inclusive_start 232.0.0.1 32 bits (4B) inclusive_end 232.0.0.1 32 bits (4B) exclusions 0 8 bits mac24_base 038000 24 bits mcast_prefix_length 11000 (24 bits) 5 bits In the example shown in Table 6.7, allocation of an additional SSM address to a MAC24 label may be performed by the Feeder dynamically upon a new request and then signalling by the additional inclusion section of the MMT2 sent to the RCSTs. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.4.5 Address mapping for non-IPv4 addresses | Many protocols also use L2 multicast apart from IPv4 and IPv6. A method may be provided to support non-IP multicast. This could be done by mapping a non-IP multicast L3 address to a L2 address, or by mapping between the LAN MAC address and the MAC24 label. The MMT2 supports non-IP multicast services, with the mappings identified by the use of their allocated protocol type field within the table. Dynamic methods are not specified and will rely on definition of an agent and a satellite control protocol between the agent and the NCC (e.g. an adapted IGMP/MLD or PIM-SM stack). |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.4.6 Address-specific issues | The IETF specifies the use of IP multicast addresses. The currently allocated set of multicast addresses by the Internet Assigned Numbers Agency (IANA) was summarized in [i.20]. This also provides general guidance on the use of the multicast address space and defines the procedures for address allocation within the multicast address blocks. With the exception of some reserved addresses the allocation of an IPv4 multicast addresses to groups is dynamic. Well-known multicast sources may be allocated a fixed and advertised multicast address. Specific multicast addresses have been statically allocated to certain roles, especially when these relate to specific protocols. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 44 Figure 6.10 shows the IANA-allocated multicast address blocks from the perspective of a satellite network. IP multicast address allocation within a satellite network has to be carefully assigned by the SNOs and SVNOs. The assignment of multicast domains and RPs has to be performed by the SNO as well. In Figure 6.9, SVN 0xFF is dedicated to globally-assigned multicast using the GLOP Block and the ADHOC Block III. These addresses are globally unique multicast addresses assigned by network operators. This will be used by the SNO to avoid simulcast of the globally unique multicast amongst the SVNOs in the SNO network. In this example, the SVN 0x0100 is also allocated a shared SVN, with MAC24 block dedicated for all other multicast traffic. This is assigned the start address of 0xFE. This can be used by the SVNO for shared multicast distribution. This avoids the need to simulcast the same content in different SVNs. The local network control block of multicast addresses (224.0.0.0 – 224.0.0.255) needs to be assigned to a preconfigured block of multicast MAC24 labels for each SVNO. These traffic flows should be independent per each IP network. Allocations have to performed to ensure that multiple SVNs are mapped to the appropriate domain, i.e. local network control packets has to be delivered between multiple SVNs if the SVNs are in the same domain. Actual allocations do not need to be for entire multicast ranges, and do not need to use a SVN-MASK of 8 bits, as used in these examples shown in Tables 6.11, 6.12 and 6.13. Figure 6.9: IANA allocated Multicast Address Blocks and their example mapping to SVNO/SVNs ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 45 Table 6.8: MMT2 example for SVN 0x0100 using shared transmission MMT2 Field Value Length svn_number 0x0100 16 bits pt_count 1 8 bits protocol_type 0x800 16 bits address_size 0x04 (4 Bytes) 8 bits mapping _section 3 8 bits inclusive_start 224.0.0.1 32 bits (4B) inclusive_end 224.0.0.255 32 bits (4B) Exclusions 0 8 bits mac24_base 017F00 24 bits mcast_prefix_length 10000 (16 bits) 5 bits inclusive_start 224.0.1.0 32 bits (4B) inclusive_end 232.255.255.255 32 bits (4B) Exclusions 0 8 bits mac24_base FE0000 24 bits mcast_prefix_length 01000 (8 bits) 5 bits inclusive_start 233.0.0.1 32 bits (4B) inclusive_end 233.255.255.255 32 bits (4B) Exclusions 0 8 bits mac24_base FF0000 24 bits mcast_prefix_length 01000 (8 bits) 5 bits Table 6.9: MMT2 example for SVN 0x0200 using shared global allocations and SVN-local address for other multicast traffic MMT2 Field Value Length svn_number 0x0200 16 bits pt_count 1 8 bits protocol_type 0x800 16 bits address_size 0x04 (4 Bytes) 8 bits mapping _section 3 8 bits inclusive_start 224.0.0.1 32 bits (4B) inclusive_end 224.0.0.255 32 bits (4B) exclusions 0 8 bits rcs_mac_base 027F00 24 bits mcast_prefix_length 10000 (16 bits) 5 bits inclusive_start 224.0.1.0 32 bits (4B) inclusive_end 232.255.255.255 32 bits (4B) exclusions 0 8 bits rcs_mac_base FE0000 24 bits mcast_prefix_length 01010 (10 bits) 5 bits inclusive_start 233.0.0.1 32 bits (4B) inclusive_end 233.255.255.255 32 bits (4B) exclusions 0 8 bits rcs_mac_base FF0000 24 bits mcast_prefix_length 01000 (8 bits) 5 bits ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 46 Table 6.10: MMT2 example for SVN 0x0300 using shared global allocations MMT2 Field Value Length svn_number 0x0300 16 bits pt_count 1 8 bits protocol_type 0x800 16 bits address_size 0x04 (4Bytes) 8 bits mapping _section 3 8 bits inclusive_start 224.0.0.1 32 bits (4B) inclusive_end 224.0.0.255 32 bits (4B) exclusions 0 8 bits mac24_base 037F00 24 bits mcast_prefix_length 10000 (16 bits) 5 bits inclusive_start 224.0.1.0 32 bits (4B) inclusive_end 232.255.255.255 32 bits (4B) exclusions 0 8 bits mac24_base FE0000 24 bits mcast_prefix_length 01000 (8 bits) 5 bits inclusive_start 233.0.0.1 32 bits (4B) inclusive_end 233.255.255.255 32 bits (4B) exclusions 0 8 bits mac24_base FF0000 24 bits mcast_prefix_length 01000 (8 bits) 5 bits |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.4.7 Source-specific multicast support with MMT2 | Currently, there is no defined syntax to support the MMT2 describing the source address for a group. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.5 Multicast management for DVB-RCS2 | The control for delivery of IPv4 multicast in static mode with active forwarding at the RCST is provided using an IGMP MIB at the RCST. This is the default case for DVB-RCS2. A candidate IGMP MIB for a RCS2 network is defined in Tables 6.11 and 6.12. The RCST should show, for debugging purposes, information about the multicast sessions it is subscribed, that is, host information to be reported to the querier when using IGMP. Management of multicast requires visibility of the active address mappings and also the filters used at an RCST, since the Service Provider will need to confirm the set of active multicast groups, the mappings to L2 and the status of group membership subscription via PIM, IGMP and/or MLD. This allows an operator to determine whether a multicast outage is due to a L3 routing/RPF issue, a L2 problem, or an upstream network problem. Traffic statistics can show information about the forwarded packets over the satellite and LAN interfaces. A multicast forwarding table defines the configuration items in the RCST, in case they are not indicated upon logon procedure in the HLS support descriptor. This applies to static modes, where the installer or the SVNO configures the forwarding mode of the RCST. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.5.1 Multicast configuration and monitoring in RCST MIB | The MIB for DVB-RCS may be organized in two tables: the Interface and Cache tables (Tables 6.11 and 6.12). The IGMP Interface table contains entries for each interface that supports IGMP on a device. For the Gateway, this includes the Core-Network and satellite interfaces, while for the RCST, the satellite and LAN interfaces. The IGMP Cache table contains one row per each IP Multicast Group for which there are active members on a given interface. Active membership should only exist on the RCST LAN interface. However, active membership may exist on both the network side and satellite interfaces of the Gateway. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 47 Table 6.11: IGMP interface table in RCST RCST Active igmpInterfaceTable Upstream-Network Side RCST LAN network igmpInterfaceIfIndex Not-accessible, assigned interface number Not-accessible, assigned interface number. igmpInterfaceQueryInterval Read-only, the RCST should not transmit queries upstream, return 0. Read-create, min = 0, max = (2^32-1), default = 125 igmpInterfaceStatus Should be enabled on both interface for all DVB-RCS RCST interfaces igmpInterfaceVersion Should be version 2 for all DVB-RCS RCST interfaces igmpInterfaceQuerier Read-only, Should be the address of an upstream IGMP Querier device for both active and passive RCSTs. Read-only, active RCSTs may report it as the satellite interface value. However, active RCSTs that participate in IGMP Querier negotiation on the RCST LAN interface may report it as a different RCST LAN device. igmpInterfaceQueryMaxResponseTime n/a, read-only, return value of 0 Read-only, value derived from observation of queries received from an upstream querier. igmpInterfaceQuerierUpTime Read-only igmpInterfaceQuerierExpiryTime n/a, read-only, return 0 Read-only, RCST may only be the querier on the RCST LAN igmpInterfaceVersion1QuerierTimer Read-only igmpInerfaceWrongVersionQuerier Read-only, the number of non-v2 queries received on this interface. igmpInterfaceJoins n/a, read only, return 0 Read-only, group membership defined to only exist on the RCST LAN. igmpInterfaceProxyIfIndex Read-only, return 0 Read-only, return a ifIndex for satellite-link interface igmpInterfaceGroups n/a, read only, return 0 Read-only, group membership is defined to exist on the RCST LAN interface. igmpInterfaceRobustness Read-create, min = 1, max = (2^32 – 1), default = 2. igmpInterfaceLastMembQueryIntv1 n/a, read-only, return 0 Read-create, min = 0, max = 255, default = 100 Table 6.12: IGMP cache table at RCST igmpCacheTable igmpCacheAddress Not-accessible (index), report the address of active IP Multicast on the RCST LAN interface. igmpCacheIfIndex Should only apt to RCST LAN interface. igmpCacheSelf Read-create, implementation specific. If RCST configured to be member of group, then membership reports are sent with the RCST's IP address but SHOULD ONLY be sent in proxy for active sessions. If the RCST is not configured to be a member, then the source IP address of membership reports SHOULD be set to the current value of the igmpCacheLastReporter address. igmpCacheLastReporter Should only apply to last reporter on RCST LAN interface. igmpCacheUpTime Read-only, Should only apply to duration of membership on RCST LAN interface. igmpCacheExpiryTime Read-only, Should only apply to duration of membership on RCST LAN interface. igmpCacheStatus Read-create. Should only apply to membership on RCST LAN interface. igmpCacheVersion1HostTimer n/a, read-only, return 0 Read-only ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 48 |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.5.2 Multicast forwarding management | New multicastFiltersTable (see Table 6.13) is needed for including the Layer 2 filters that will be used by the ST to receive the multicast streams and forward them to the user LANs/VLANs. This table is unique for all the SVNs present in the ST. According to the value of parameter vrfMulticastMappingMethod, the mechanism for the population of this table differs. In case the MMT2 method is used, the table is automatically composed and updated by the ST upon MMT2 decoding. Each IP multicast flow should be delivered to the corresponding SVN interface, only when their respective membership group is active. The SVN interface can be deducted from the respective SVN number. Table 6.13: Multicast Filter MIB table Element Range Description in HLS Changes for HLS multicastFiltersTable SEQUENCE OF multicastFiltersTableEntry - New table multicastFiltersTableEntry SEQUENCE { multicastFilterIndex, multicastFilterSVNnumber, multicastFilterRCSMAC, multicastFilterInclStart, multicastFilterInclEnd, multicastFilterExclStart, multicastFilterExclEnd, multicastFilterStatusRow } - multicastFilterIndex INTEGER - Table index of the multicast entry. multicastFilterSVNnumber INTEGER SVN where the multicast flow should be delivered. This is a link to the virtual interface in vrfGroupTable. multicastFilterRCSMAC OCTET STRING - MAC24 of the multicast group. multicastFilterInclStart InetAddress - First multicast IP address included in the range. multicastFilterInclEnd InetAddress - Last multicast IP address included in the range. multicastFilterExclStart InetAddress - First multicast address excluded in the range. multicastFilterExclEnd InetAddress - Last multicast address excluded in the range. multicastFilterStatusRow Row Status - The row status, used according to row creation and removal conventions. A row entry cannot be modified when the status is marked as active(1). A row can be created either by createAndGo and automatically change to active state or createAndWait to add more parameters before becoming active. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 6.5.3 Multicast statistics | The statistics for transmitted and received multicast packets can be obtained from the interfaces MIB group, in parameters ifInNUcastPkts and ifOutNUcastPkts, by locating the corresponding SVN interface (identified by parameters ifIndex and ifPhysAddress). |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 7 QoS support | This clause presents an overview of the Differentiated Services (DiffServ) Quality of Service (QoS) model for DVB- RCS2 transmission systems together with implementation guidelines regarding its use. It briefly describes the current IP DiffServ model in terrestrial networks, and describes this model for DVB-RCS2 systems. Finally, it provides examples of QoS configuration for a range of terminal profiles. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 49 The Higher Layers (HL) support at the RCST contains the relevant components to implement QoS support on the Return Link (RL). This includes traffic classification, policing functions, and scheduling according to the HL service associated with traffic flows. The RCST QoS model in DVB-RCS2 HLS [i.4] aims to satisfy the capacity requirements for different users and services. The four basic components of the QoS model in DVB-RCS2 HLS are: • RCS2 satellite terminal, RCST. • Network Control Centre (NCC) and Network Management Centre (NMC). The NCC controls the interactive network (control plane). The NCC is in charge of element and network management functions (in the management plane). • RCS2 Gateway (GW). • Operations Support System (OSS). Figure 7.1: Planes in the higher layers of DVB-RCS2 HLS standard The RCST SW may be customized for a given terminal profile. Although, it may also be compatible across multiple terminal profiles. The main DVB-RCS2 terminal profiles are: • Consumer / SOHO • Institutional / Corporate • Backhauling • Multi-dwelling • SCADA Table 7.1 provides an example of the functional QoS requirements that can be provided for different terminal profiles: ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 50 Table 7.1: QoS requirements per terminal profile Consumer/ SOHO Institutional/ Corporate Backhauling Multi-dwelling SCADA Number of IP QoS classes for HL Service Up to 5 5-7 1-3 (GSM) 3-5 (maritime, in- flight, on train) 5 1-2 Traffic profiles for transmission and reception Asymmetric (consumer) A/Symmetric (SOHO) Symmetric Low-latency Asymmetric Fair sharing between users Asymmetric Bursty traffic User Services Internet, VoIP, VPN, P2P, gaming, streaming Corporate Military Surveillance Disaster Relief Internet in-flight. GSM Satellite/LTE Internet access, VoIP Monitoring of real time industrial processes. Environmental monitor The DiffServ model [i.21], [i.22] and [i.23] defines an IP QoS architecture based on packet marking. In this model, policy-based management mechanisms are used for prioritizing network resources to meet the requirements of specific traffic types on a per hop basis. No explicit signalling is used to communicate with DiffServ routers, instead a set of Traffic Classifiers (TCs) are used to assign flows to one of a set of pre-defined Behaviour Aggregates (BAs). The classification is performed by inspecting packet header fields, such as IP addresses, ports, and the Differentiated Service Code Point (DSCP) [i.21]. The DiffServ model defines consistent QoS operation within the routers that form a part of the network called the DiffServ domain. The domain consists of a contiguous set of routers operating with a common set of service provisioning policies. It is common practice to provide traffic conditioning, including admission control, shaping and policing at the edge of a DiffServ domain [i.24]. The DiffServ framework for policy-based admission control [i.25] describes the various components that participate in policy decision-making (i.e. Policy Decision Point, PDP, or Policy Enforcement Point, PEP). Traffic conditioning of admitted traffic may be performed using "meters" to measure the properties of each BA [i.23], [i.26] and [i.27] against a traffic class (or traffic specification). A Policy Enforcement Point may police the PDUs from non-conformant flows (i.e. These may be marked, dropped or shaped). The treatment of the traffic forming a BA is characterized by a Per Hop Behaviour (PHB) [i.21] and [i.24]. Within a DiffServ domain, the network operators may choose to support any combination of standard or operator- specific PHBs. The current set of standard PHBs defined by the IETF is: • Expedited Forwarding (EF) [i.28] and [i.29] • Assured Forwarding (AF) [i.30] • Default (Best Effort) [i.21] Each standard PHB has been assigned a standard DSCP. EF has been assigned DSCP 46 and BE has been assigned 0. The DSCPs assigned for the AF PHB group are given in Figure 7.2. Figure 7.2: PHB assignment to DSCP for AF PHB group Other PHBs may also be standardized and vendors/operators may also introduce their own PHBs. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 51 |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 7.1 QoS Model in DVB-RCS2 | The DVB-RCS2 specification adopts the DiffServ model. Each RCST uses a set of TCs specified in a configured policy(class map) to map packets received on the LAN Interface to a specific HL Service (see clause 7.1.3.1). The set of classified packets handled by a HL Service form a BA. Traffic conditioning of admitted traffic may optionally be performed. All packets assigned to a BA receive the same treatment (i.e. the same HL Service, that is, they are assigned to the same queuing and IP scheduler treatment). This treatment is characterized by a PHB. Within the RCST, each HL Service is mapped to a Lower Layer Service (LL Service) and a set of Request Classes (RCs) in the control plane. The following table provides examples of the relationships between a set of IP service classes and applications, based on IETF recommendations [i.31]. Table 7.2: PHB – Example application mapping Service Class Name DSCP Name Application Examples Network Control CS6 Network routing Telephony EF IP Telephony bearer Signalling CS5 Telephony signalling Multimedia Conferencing AF41, AF42, AF43 H.323/V2 Video conferencing Real Time Interactive CS 4 Video conferencing and Interactive gaming Multimedia Streaming AF31, AF32, AF33 Streaming video and audio on demand Broadcast Video CS3 Broadcast TV & live events Low-Latency Data AF21, AF22, AF23 Client/Server transactions Web-based ordering OAM CS2 OAM&P High-Throughput Data AF11, AF12, AF13 Store and forward applications Standard DF (CS0) Undifferentiated applications Low-Priority CS1 Any flow that has no BW assurance Considering Table 7.2 and the terminal profile requirements (in Table 7.1), Table 7.3 can be used to identify a set of BAs that would be appropriate for a specific deployment scenario. Table 7.3: Behaviour Aggregate – Example terminal profile mapping Scenario Behaviour Aggregates Consumer/ SOHO EF, (AF31), (AF21), (AF11), DF Government/ Corporate EF, AF31, AF32, AF21, AF22, AF11, DF Backhauling EF, AF21, (AF31), (AF41), DF Multi-dewlling EF, AF31, AF21, AF11, DF SCADA AF11, BE The remainder of this clause identifies the main components of the RCST QoS architecture and reviews the relationship between key functions. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 7.1.1 RCST2 Connectivity Aggregate and Connectivity Channels | Traffic received from the LAN Interface that is to be forwarded by an RCST is divided into one or more Connectivity Aggregates (CA), based on the next hop layer 2 destination to which it is to be forwarded. The term "aggregate" is used generally in the HL user plane to indicate a sequence of HL satellite protocol data units (HLS PDUs). The CA is, hence, the output of a Layer 3 routing or Layer 2 forwarding decision and reflects the interface on which the traffic will be carried over the satellite network (see Figure 7.3). The following examples illustrate the CA concept: • In a star network, an RCST could use a single CA to forward all traffic towards the GW. • In a mesh network, an RCST may configure multiple CAs; one could offer connectivity to the GW, and others could offer direct connectivity using mesh connections to other RCSTs. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 52 Figure 7.3: DVB-RCS2 Routing/Forwarding Functions Each allocated timeslot is associated with a specific connectivity channel, and hence a single CA. The choice of how many CAs are used depends on how the traffic is to be managed and whether allocated time slots may be used to reach multiple destinations. Each CA requires a separate instantiation of the QoS framework (i.e. requires an independent set of HLS PDU Queues, a corresponding set of HL Services, set of independently managed QoS and RRM entities, etc.) and supports one Connectivity Channel (CC), which is a physical stream of transmission of bursts. Traffic is routed to a particular CA as a result of a routing decision to a next hop address. Alternatively, a single routing entry could direct traffic to one CA, which later maps traffic to one or more Link Streams. In the second approach, a single set of HL and LL entities may be instantiated. A CA may utilise multiple Layer 2 streams. Several possibilities exist, for example: • A single CA does not necessarily imply a single link-layer destination (next hop MAC24 address) or a single Link Stream; this is because DVB-RCS2 allows encapsulating multiple PPDUs with ALPDUs in the same time slot even though these may be destined to different L2 addresses. • A Link Stream may be used to allow an RCST to identify a mesh destination in a DVB-RCS2 network that supports this connectivity, e.g. when more than one destination is reachable via a CC. The connectivity offered in a mesh system may demand that multiple CAs are used. • When a CA is used with a multiple access link, it is envisaged that one Link Stream could be configured for each L2 destination. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 7.1.2 RCST QoS Services | QoS Services are realized in an RCST using a combination of HL and LL Services. CAs are typically subdivided (classified) by a set of TCs that assign the traffic to a specific BA. The TC information may also contain meta- information regarding the traffic specifications for the BA, such as peak-rate, sustainable rate, etc. These values can be used for traffic conditioning (as a DiffServ Policy Enforcement Point) when supported by an RCST. The traffic forming a BA is queued in an HLS PDU queue, which is then mapped to a LL Service Aggregate (SA). After HL and LL processing, the CA will be finally transmitted using a CC. Figure 7.4 presents an example using rectangles to represent functional entities and octagons to represent selection functions. The scheduler (represented by an oval) is an abstract function that determines how HL PDUs are mapped to a SA. Control functions relationships are represented by dashed lines and data flow by solid lines. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 53 Figure 7.4: DVB-RCS2 QoS Components A more detailed explanation of the operation of the user and control planes is provided in the next clauses. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 7.1.2.1 User plane QoS | This clause describes QoS processing by an RCST for transmission on the RL. Each RCST has at least one CA that it uses for transmission to the GW. An RCST may also create CAs for other destination within the satellite network (e.g. when supporting mesh connectivity). Each PDU belonging to a CA is assigned to a single BA, based on a packet classifier that matches the packets to one TC. A TC is implemented as a set of records in the IP classification table (see clause 7.3); each TC matches a set of fields in the IP or L2 header. A classification rule may be as simple as matching only the DSCP or may be more complicated, e.g. involving matching several IP fields with a multi-field (MF)-classifier. A packet classifier may use multiple fields to form a TC: • Layer 2 information may be used as part of this classification. For instance, a policy may be configured to associate an 802.1pQ PCP value with a specific BA, or a classification rule could be written to assign L2 packets (e.g. LAN control information) to a specific BA based on the Ethertype. This classification may be applied to non-IP traffic. • At Layer 3, an IP packet may be classified based on the DSCP markings and other IP header information. Together, these fields may be used to select the BA. IP traffic with the Type Of Service field not in line with DiffServ semantics may use the Class of Service (CoS) semantics, rather than those specified by the DiffServ architecture. • At Layer 4, deep packet inspection may match the port information and other IP payload information to assign the packet to a BA. Since different levels of classification may result in assignment to a different BA, the RCST needs to specify which fields to trust when there is conflicting information in the TCs. When a RCST acts as a DiffServ Policy Enforcement Point, the TC may also specify flow properties (e.g. traffic average rate, max burst size, etc.). These properties are used to decide whether the rate of a flow is conformant or non- conformant to traffic specifications. A PDU belonging to a non-conformant flow may be marked (changing the DSCP and/or Explicit Congestion Notification value), and/or re-queued to a different BA, or dropped (discarded). This use implies that a TC may be associated with an additional BA to be used for non-conformant traffic. Once PDUs are assigned to a BA, they receive the same queuing and IP scheduling treatment (i.e. they are assigned to a single HLS PDU queue). Each BA is characterized by a specific queuing strategy and scheduling method. The traffic forming a BA should be sent using one SA. This set of attributes is collectively referred to as a Higher Layer Service in the control plane. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 54 An SA comprises the PDUs from one or more BAs, and associates these with a priority/precedence. All PDUs assigned to the same SA receive the same treatment by the LL Service. Within the LL, a SA is transported using a Link Stream (LS) that carries a sequence of L2 packets. For example, a LS may be associated with Payload-adapted PDUs (PPDUs) of a LL logical flow. Packets are finally multiplexed into bursts or Transmission Streams (TX Streams) for transmission over the air interface. The precedence of a LL service is used to inform scheduling decisions when a transmission opportunity is made available to an RCST. When more than one SA is defined L2 pre-emption may be used. This allows the QoS system to suspend transmission of a partially-transmitted PDU (Link Stream packet) from a lower priority SA at the end of a transmission burst, and initiate (pre-empt) the next transmission burst with a PDU from a higher priority SA. The transmission of the lower priority SA is resumed in a later timeslot. This method can be used to upperbound the jitter experienced by higher priority SA traffic. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 7.1.2.2 Control plane QoS | The HL Service is defined as a per-hop treatment of Layer 3 PDUs characterized by a PHB. This is a management construct that includes the policy needed to instantiate the PHB and relate this to an HLS PDU queue. Each HL Service corresponds to one BA. This defines the parameters that are needed to support PHB-specific operations, including queuing and scheduling, and the SA to be used. A single PHB may be instantiated to form multiple instances of the HL Service. Conversely, a single HL Service may be used to support multiple PHBs using a single BA. However, since the HL Service can not differentiate the treatment of PDUs within the BA, a set of HL Services need to be instantiated to realize DiffServ QoS. A consistent QoS treatment across HL and LL is guaranteed by defining one LL Service (SA) for one or more HL Services. Each LL Service is created by the LL Service Descriptor in DVB-RCS2 Lower Layer Specification. The LL Service provides an interface to access the satellite resources. The configuration of the LL services associated to a SA and its corresponding Link Streams determines: • the allowed mapping between Link Streams and Request Classes (RCs), • the allowed mapping between Link Streams and dedicated-access allocation channels, • the allowed mapping between Link Streams and random-access allocation channels. An LL Service specifies the types of Allocation Channels (AC) that may be used for each SA. The AC identifies a portion of the RL capacity that is available for use by one or more LSs: • The Dedicated Access AC (DA-AC) receives allocation by means of explicit demand/assignment methods or free capacity assignment (FCA). • The Random Access AC (RA-AC) represents a portion of the return link spectrum that is offered for random access for multiple terminals. NCC may use a load-control algorithm to control the level of contention on the RA channel. An LL Service may allow an SA to use one or more (DA-ACs) LL Service or one or more RA-ACs. In addition, it may optionally be mapped to other AC, with each AC mapped to a connectivity channel. The LL Service can also inhibit access to the DA-AC or RA-AC, for instance because an RCST does not support these LL Services. An RCST uses an AC to select a specific RC or for QoS differentiation: • The LL Service contains a reference to the AC and the RC. This defines implicitly an association between the AC and the RC. When the RCST generates a Capacity Request (CR), it inserts the RC identifier (RC_index) into the CR message to communicate this value to the NCC. • An AC is also used to differentiate connectivity channels, when multiple capacity categories are used. The NCC is responsible for distribution of DA-ACs and RA-ACs. The NCC can address a specific DA-AC by inserting an Assignment_ID in the TBTP2. • Each DA service is the control plane correspondence for the user plane DA-AC. This is specified in the RCST configuration. As seen from an RCST, there is a one-to-one correspondence between a DA service and a DA allocation channel. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 55 • The RA service is the control plane correspondence for the user plane RA-AC. This is defined by the resources provided to the associated RA-AC (as controlled by the NCC), the RA Load Control parameters associated with the RA-AC and the current loading of the RA-AC by the RCST. An RCST assumes that a DA Service will be assigned capacity by the NCC as specified for the nominal RC associated with the DA Service. The DA Service specification can then be inferred from the configuration of this RC. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 7.1.2.3 Management plane QoS | The HL Service mapping contains a number of managed QoS parameters that characterize the HL Service, such as: MinRate, MaxRate, MaxIngressBurst, MinIngressBurst, MaxDelay, MaxLatency and LinkRetransmissionAllowed. Also, it contains information relating to the queue behaviour, such as SchedulingType. Table 7.3 in [i.1] provides the minimum set of HL Service parameters. Additional parameters could be added by the implementor, if necessary, to better specify the expected QoS behaviour of the user PDU within the RCST. In addition, in Table 7.4, some examples are provided for queue configuration for some BA / HL Service. Table 7.4: Example HL Service Configuration HL Service (PHB) Dropping Mechanism Scheduling Type Queue size EF Tail Drop FIFO MaxIngressBurst*MaxDelay AFxx Random Early Detection WFQ MaxIngressBurst*MaxDelay BE Tail Drop FIFO MaxIngressBurst*MaxDelay Traffic is classified by matching against a set of TCs, an example is shown in Table 7.5. Table 7.5: Example TC Service Configuration MF Classifier HL Service Metering System EF DSCP (46) EF Disable AFxx DSCP AFyy Single Rate Three Colour Marker; Two Rate Three Colour Marker BE DSCP (0) BE Disable |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 7.2 QoS organization configuration | The organization of BAs and Link Streams and the distinction of LL and HL Services provide a number of ways to configure the QoS support in a RCST. Two distinct mappings can be identified for a star system: • A mapping may regard all allocated timeslots as belonging to a single SA. This is the simplest method. It queues packets by BA within the HL, and requests capacity using one or more RCs. Although different policies may be used to request capacity for different TCs, all allocated timeslots are used as one service by the scheduler, which optimizes use according to the assigned LL Service. • A mapping may provide a strict separation between a set of LL Services. This queues packets by BA within the HL, and also may request capacity using more than one RC. Different policies may be used to request capacity for different TCs. The allocated timeslots are differentiated at the scheduler by LL Service, which seeks to assign the traffic to the allocations made in response to each RC. A policy may be used to reassign unused timeslots to other traffic. Figure 7.5 illustrates two organizations of the QoS system in a RCST. DVB-RCS2 does not specify a particular method for scheduling. In this example, a scheduler is assumed to be triggered by allocation of a timeslot. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 56 BA1 SA1 (a) Differentiation of services HLS1 LLS1 BA2 SA2 HLS2 LLS2 (b) LLS aggregation & pre-emption BA1 SA HLS1 BA2 HLS2 LLS1 LLS2 BA1 SA1 HLS1 BA2 SA2 HLS2 LLS (c) LLS differentiation & multiplexing (d) QoS aggregation BA1 SA HLS1 BA2 HLS2 LLS Figure 7.5: Example QoS mappings An example of the first organization is shown in Figure 7.5(a). Two BAs are each mapped to a SA. The HL Service associated with each BA is also mapped to a distinct LL Service. This organization can be used to support two distinct services that operate independently. Each LL Service could be independently billed, policed, and cannot be adversely impacted by traffic assigned to another LL Service. Independent L2 allocation methods are used to request transmission resource for each BA, and the NCC using a corresponding assignment_id identifies the allocated timeslots. The RCST scheduler will use the assignment_id to schedule SA traffic. A different organization is achieved when the BAs are mapped to separate SAs, but their respective HL Service is multiplexed to a common LL Service (Figure 7.5(b)). This organization allows the scheduler to use the optimum policy to schedule the use of the allocated capacity (according to the parameters set in the HL Service for each BA. Since different Link Streams are used at L2, the L2 scheduler is responsible for determining the order of transmission of packets and may support pre-emption of lower priority SAs. Other organizations are also possible using the RCST QoS architecture. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 7.2.1 Scheduling in RCST | DVB-RCS2 intentionally does not specify the semantics of scheduling in RCST. This leaves implementers the flexibility to perform scheduling decisions either in the lower layers, the higher layers, or a combination of the two. In the following example, it is assumed that a scheduler is used that is triggered by a transmission opportunity on either a RA or DA channel. This is most easily envisaged in the DA case, where the task of the scheduler may be to determine whether there is any data that should be sent in the transmission opportunity, and if so what data should be sent. The data to be sent comprises: • Partially sent HLS PDUs pending completion of the final fragment(s) • HLS PDUs queued at the higher layer If two BAs are mapped to the same SA, scheduling decisions are made before packets are encapsulated at L2 (Figure 7.5 (c) and (d)). However, distinct allocation methods can still be used for the two BAs. This organization allows allocation of resources to different streams of traffic (e.g. VoIP, web traffic, interactive services, etc.) within the same allocation pool. In this case, L2 pre-emption may not be required. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 7.2.2 Example use of RCST QoS system model | Figure 7.6 illustrates the relationship between modules the higher layer QoS functions and the lower layers QoS functions. The diagram is intended to be informative and does not mandate any particular internal structure of an RCST. Solid lines represent the flow of PDUs and other data through the system, whereas dashed lines are used to denote control relationships. Simple functions or objects are represented by boxes, selector mechanisms by hexagons, and complex objects by pentagons. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 57 Figure 7.6: Logical HLS QoS Processing In the diagram, the data paths are represented by dashed lines and control paths by dashed lines. Traffic arriving at the LAN network interface of an RCST has been divided into several Traffic Classes (TCs). These classes are mapped to 5 per-hop behaviours (PHBs). These traffic classes may for instance reflect a best effort Diffserv Code Point (TC1), and unknown service category (TC2) – in this case mapped to the Best Effort (BE) PHB, an Assured Forwarding codepoint mapped to one of the two AF PHBs, and an Expedited Forwarding class mapped to the EF PHB. The final traffic class maps to be a special-purpose class, the XX PHB. Each HLS PDU queue (behaviour aggregate) is in turn mapped to a Link Stream (service aggregate) for transmission (ST1-ST3). The Radio Resource Management (RRM) object is responsible for requesting capacity from the NCC. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 58 The outputs of the HLS PDU Queues hold the data to be sent over the lower layer service. This implies the action of an IP scheduler (represented by a white oval). This may be understood to be activated each transmission opportunity (notified by the TBTP2) to select the PDUs that are segmented into the stream. The selection is based on the PHBs (which indicate the lower service), and link-layer information. This ensures that PDUs or segmented PDUs are sent using the corresponding allocation channel. When required, PDUs pass through a segmentation function, so that any unsent data is postponed to a later scheduling opportunity. Each segment is then encapsulated into one of the configured streams (ST1-ST3 in the diagram) and is then placed in the burst for transmission. The scheduler could use a strict priority scheduler or a weighted priority scheduler, but is not specified in the present document. Since in this example there are three Link Streams, ST1 can preempt ST2 or ST3. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 7.3 QoS configuration management | The RCST basic QoS configuration may be provided in the new configuration file download or via TIMu NLID messages. It may also be provided directly by the installer (e.g. by manually configuration or via local configuration file download). The first time that the RCST enters the system, it is recommended to always verify the full HLS configured data. The exact value of the QoS HL parameters will depend on the RCST profile and should follow the recommended values provided in previous clauses. The NMC side could have different QoS templates depending on the RCST profile in the system. This information should be part of the RCST commissioning together with other HL parameters. Figure 7.7: QoS configuration management The HLS QoS configuration should, at least, contain the following information: • IP Classification table: This defines a TC in the form of a table that maps each PDU to a specific BA,. If there is no entry in this table, then there is no way to classify the traffic, and the RCST by default may drop the user traffic. The exact number of entries in IP classification table will be system dependent. However, at least one default entry should be provided (e.g. to a best-effort BA). • HL Service mapping table: This maps the HL services to one LL service for consistent QoS treatment. This table should contain at least one HL service per LL service provided during logon messages. The setting of NMC QoS parameters per RCST is given by the SVNO. The SVNO is responsible for the traffic functions, IP routing, and QoS. Therefore the SVNO should provide the RCST QoS HLS templates per profile in the NMC. The NMC should contain different RCST profiles depending on the terminal profile type of RCST that the SVNO works with. It may be possible to have more than one template per profile, which will be a system implementation decision. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 59 7.4 QoS management and control in regenerative mesh networks |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 7.4.1 DVB-RCS2 Logon with regenerative mesh support | Dynamic connectivity support, connection control protocol support and version, and transparent mesh capabilities are included in the LL capabilities and HL capabilities groups of the logon element types sent by the RCST. The Logon TIMu includes the Logon Response Descriptor, which may provide one dedicated access allocation channel (DA-AC) for control and management, providing the resources for the mesh signalling connection. This is done by associating an Assignment ID to signalling. More information to establish the signalling connection may be provided in the NLID descriptor. The Logon TIMu also includes the Lower Layer Service Descriptor, where the allocation channel applying to each lower layer service, and its corresponding RC (Request Class), are indicated. For multi-beam mesh systems, a different allocation channel per physical destination may be needed. This is due to the possibility that the physical resources associated to the different destination are disjoint. Such is the situation when the satellite switching is performed at Layer 1. In this case, different Assignment IDs can be assigned per destination beam supported by the RCST. The RC associated to each AC can be independently configured, according to the LL Services configuration. Configuration of default values for the HL QoS tables (IP Traffic Classification and HL Service tables) might take place by reception of NLID descriptor (in the form of SNMP set commands) included in the Logon TIMu. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 7.4.2 HLS Maintenance | The SVNO may (re)configure or add new entries in the IP Traffic Classification and HL Service tables of the RCST, using SNMP protocol or other IP-based method (e.g. configuration file download). Certain parameters of the HL Service table entries will not be modifiable by the SVNO, as the LL service associated to an HL Service. The default entry of the IP Traffic Classification and HL Service tables should not be modified or deleted by the SVNO. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 7.4.3 QoS Configuration for regenerative mesh systems | The RCST QoS data structures used for regenerative mesh systems are: 1) Traffic classification and services tables (): - IP Classification Table (defined in [i.1]). - HL Service Mapping Table (defined in [i.1]). - LL Service Table, created after reception of the LL Service descriptor at logon. - RC Table, created after reception of the LL Service descriptor at logon. 2) DCP specific tables: - Active Connection Table. - Connections timeout value. This value may also be established by using DCP. The Active Connection Table (see Table 7.6) lists the active DCP connections and needs to include the following parameters: ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 60 Table 7.6: DCP Active Connections Table Parameter Description ActiveCnxIndex The index in the DCP active connections table. ActiveCnxRefId A reference for the DCP connection. ActiveCnxMACSrcAddr MAC24 address of the originating interface in the source RCST. ActiveCnxMACDestAddr MAC24 address of the destination RCST reception interface, or multicast MAC24 for unidirectional multicast connections. ActiveCnxType Unidirectional (unicast or multicast) or bidirectional. ActiveCnxService QoS Service for the connection (used for transmission). ActiveCnxAssignmentId Reference to map TBTP2 resources assigned to the connection for this RCST. ActiveCnxOtherAssignmentId Reference to map TBTP2 resources assigned to the connection for the peer RCST. ActiveCnxIPv4SrcAddress In the source RCST, IPv4 address of the LAN interface that received the IP traffic packet triggering the connection. ActiveCnxIPv4DstAddress IPv4 address of the peer RCST for this connection, taken from the Connection Establishment Request message. ActiveCnxIPv6SrcAddress In the source RCST, IPv6 address of the LAN interface that received the IP traffic packet triggering the connection. ActiveCnxIPv6DstAddress IPv6 address of the peer RCST for this connection, taken from the Connection Establishment Request message. |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 7.4.4 QoS MIB Objects for regenerative mesh | The QoS parameters, that determine the QoS profile and Allocation Channel (AC) corresponding to one mesh connection, can be extracted from the HLS tables (IP Classification and HL Service MIB tables), included in the dvbRcs2QoSConfiguration MIB group [i.1] and the LL Service Descriptor (LL Service and RC MIB tables). Table 7.2 in [i.1] includes the defined IP TCs. This table links to the HL services table through parameter IPClassHLSAssociation. The TC entry may apply to only one satellite SVN, if its associated HL service is defined for a specific interface. The RCST may discard VLAN frames if their user priority does not match the value specified by IPClassVlanPri parameter. Also, the RCST can drop IP packets based on IP header values through parameter IPClassAction. In this way it can be avoided that a user packet triggers a DCP connection request, the table entry is then acting as a reverse firewall. Table 7.3 in [i.1] characterizes HL services and links them with LL services. Each entry of the HL Service table is associated to a LAN interface number of the RCST, which will map to the interfaces MIB group, being applicable only to one SVN (linked to a VLAN_ID). This means that HL Services should be replicated (when necessary) for each SVN supported by the RCST. The recommendation is that the IPClassVLANID field in the IPClassTable is left empty when the entry of the table applies to all the SVNs supported by the RCST. The VLAN mapping table (Table 8.14) maps user VLAN_IDs and satellite SVNs. Table 7.7, constructed from the LL Service Descriptor in the Logon Response, maps LL services with RCs and ACs. A reference for the LL service that the RCST intends to use for the mesh link will be included in the DCP establishment request message. This LL service maps to one RC class. Table 7.7: LL Service parameters Element Description LLserviceIndex Index of LL service Table. LLserviceRCIndex A 4 bit field indicating the nominal request class for the associated Link Service. LLserviceDAACIndex A 4 bit field indicating the nominal dedicated access allocation channel associated with the Link Stream. The Assignment ID associated to the request class has an offset to the Assignment ID Base equal to the nominal_da_ac_index. LlserviceCD_RCmap A 16 bit field indicating the allowance to conditionally map resource demand for the associated Link Stream into capacity requests for other RCs, with bit 0 referring to rc_index=0, bit 1 referring to rc_index=1 and so on. LLserviceCS_DAACmap A 16 bit field indicating the allowance to conditionally map traffic from the Link Stream into the different dedicated assignment allocation channels, indicated by a flag for each DA- AC, with bit 0 referring to da_ac_index=0, bit 1 referring to da_ac_index=1 and so on. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 61 Table 7.8, derived from the LL Service Descriptor, defines the RCs in the system that are usable by the RCST. These RCs are used by mesh links CCs according to the QoS service selected by the RCST originating the request. Table 7.8: RC table parameters Element Description RCindex The RCST by default maps its default request class to rc_index 0 RCcontantAssignment Flag to indicate if constant non-solicited assignment is provided for the RC Values: Non-solicited(0), Solicited(1) RCvolume_allowed Flag to indicate if A/VBDC requests are allowed for the rc_index Values: NotAllowed(0), Allowed(1) RCrbdc_allowed Flag to indicate if RBDC requests are allowed for the rc_index in kilo bits per second Values: NotAllowed(0), Allowed(1) RCmax_service_rate Field that indicates the maximum service rate for the rc_index. The maximum allowed RBDC equals this level substracted by the CRA in kilo bits per second RCmin_service_rate Field that indicates the minimum rate that can be expected assigned when actively requesting any dynamic capacity for the rc_index RCconstant_service_rate 16-bit field indicating the admitted CRA level associated with the request class in kilo bits per second RCmax_backlog 8-bit field indicating the max volume request backlog that the NCC will accept to hold for the rc_index in kilo bits per second |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 8 Satellite Virtual Networks and VLANs | |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 8.1 Mapping of SVN tags to lower layer fields | This clause provides guidelines on the SVN tag mapping to lower layer fields. The concept of SVNs is explained in [i.1]. At layer 2, each logical RCST network interface towards the satellite system has a unique 24-bit MAC24 label that consists of an SVN-number or SVN-prefix and an SVN interface ID. The boundary between the two is variable and configured via L2 signalling and may be different for difference SVNs running in the same system. When using different SVN prefix lengths, care should be taken to assign addresses so that the SVN numbers are non-ambiguous. The allowed range of the prefix length is 1 to 16 bits, so that an RCST should support at least two SVNs – SVN 0 for management and one or more SVNs for user traffic. Figure 8.1: Format of MAC24 The length of the SVN number (the boundary) is signalled in the L2 Logon Response Descriptor. This descriptor can configure up to 15 RCST addresses, the SVN number length (MAC24_prefix_size) is set for each of them independently and care should be taken that no overlapping addresses are created. The prefix length field svn_prefix_size has a size of 5 bits, but its contents are restricted to the range [1..16]. The actual value should be decided based on the maximum number of RCSTs to be supported in the given SVN. If, for example, up to 2 048 RCSTs are to be supported, the interface ID size should be at least 11 bits and the svn_prefix_size should be less than or equal to 13. The Logon Response Descriptor also configures the assigned MAC24 label in the unicast_mac24 field, and the default SVN number in default_svn_number. When encapsulating a higher layer PDU, the RCST should compare the SVN number of the packet to the default_svn_number. If these match, the RCST should use a 0-byte packet label for encapsulation (i.e. label type 2). If they do not match, the RCST should take most significant byte of the MAC24 and place this into the ALPDU label with label type 0. ETSI ETSI TR 101 545-5 V1.1.1 (2014-04) 62 |
fc9c999440d0c3f544c424c8c56027ec | 101 545-5 | 8.1.1 MAC24 address assignment to terminals | The MAC24 addresses assigned to a terminal should provide non-ambiguous mapping to and from SVN/interface-ID. Figure 8.2: MAC24 assignment example The upper 8 bits of all MAC24 addresses assigned to a terminal explicitly or implicitly (multicast addresses) should be unique as in Figure 8.2. The number after the slash is the SVN-prefix length and the bold digits are the SVN-prefix. When sending a higher layer PDU to using any of the four addresses the ALPDU label (the upper 8 bits) will be different in each case. This means provided the hub receiver with the means to decide to which of the four SVNs the packet belongs. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.