Split (1)

train · 617k rows

text stringlengths 2 132k	source dict
enforcement, 810-8 Environmental documents geotechnical report, 510-5 Environmental hearing, 210-11 Environmental Review Summary (ERS), 330-7 Erosion prevention, 1350-2 ERS. See Environmental Review Summary Escape ramps, 1010-4 Evaluate upgrade (EU), 325-7 Exit pattern uniformity, 940-4 F Farmlands, 210-7 Farm road app...	{ "page_id": null, "source": 7320, "title": "from dpo" }
length, 630-2 length of grade, 630-2 maximum grades, 630-2 minimum grades, 630-2 minimum length of vertical curves, 630-2 railroad crossings, 630-4 vertical alignment, 630-1 Geometrics horizontal alignment, 620-1 HOV direct access, 1055-5 vertical alignment, 630-1 Geosynthetics, 530-1 applications, 530-2, 530-16 defini...	{ "page_id": null, "source": 7320, "title": "from dpo" }
210-12, 1430-2, 1430-12 administrative appeal, 210-15 advertising, 210-7 combined, 210-15 corridor, 210-11 design, 210-12 environmental, 210-11 examiner, 210-13 findings and order, 210-14 formal, 210-4 informal, 210-4 notice, 210-7 open format, 210-4 preparation, 210-8 requirements, 210-6 sequence, 210-16 study plan, 2...	{ "page_id": null, "source": 7320, "title": "from dpo" }
510-5 hydraulic considerations, 1210-1 safety considerations, 1210-3 I Illumination, 915-14 bridges, 840-5 construction zones and detours, 840-4 roundabouts, 915-14 Impact attenuator systems, 720-1 design criteria, 720-7 object markers, 830-8 older systems, 720-6 permanent installations, 720-1 selection, 720-7 work zon...	{ "page_id": null, "source": 7320, "title": "from dpo" }
access, 1430-1 approach, 1420-2 definitions, 1420-2, 1420-4 full control, 1430-3 level of control, 1430-1 limits, 1430-15, 1430-16, 1430-17, 1430-18, 1430-19, 1430-20, 1430-21, 1430-22 modifications to, 1430-12 modified control, 1430-8 partial control, 1430-5 references, 1420-1 vocabulary, 1420-5 Limits turnback, 1430-...	{ "page_id": null, "source": 7320, "title": "from dpo" }
National ITS Architecture, 860-1 New Jersey shape barrier, 710-15 Noise barriers, 1140-1 design, 1140-1 documentation, 1140-4 earth berm, 1140-2 geotechnical investigation, 510-8 noise wall, 1140-2 procedures, 1140-4 wall types, 1140-3 Non-NHS highways, 325-2 Nonrecoverable slope, 700-2 Nonseparated HOV lanes, 1050-2, ...	{ "page_id": null, "source": 7320, "title": "from dpo" }
1430-5, 1430-8, 1430-10, 1430-12, 1430-13 Portable changeable message signs, 810-10 Portable traffic signals, 810-12 Preliminary signal plan, 850-14 Principal arterial, 440-12 Private land R/W transactions, 1410-4 Profile grades, 440-10 Program elements, 150-6 Index Design Manual M 22-01 Page 12 May 2006 Programming R/...	{ "page_id": null, "source": 7320, "title": "from dpo" }
attenuator, 720-3, 720-5, 720-12 Recoverable slope, 700-2 Recovery area, 700-2 Regional Transportation Improvement Program (RTIP), 120-12 Regional Transportation Planning Organizations (RTPO), 120-6 (RTPO),, 120-3 Regulatory traffic control strategies, 810-8 Reinforced slopes, 1130-1 geotechnical investigation, 510-8 R...	{ "page_id": null, "source": 7320, "title": "from dpo" }
roadway widths, 640-2 widths, HOV facilities, 1050-8, 1050-11, 1050-12 Roadway sections, 640-6 Rock, investigation, 510-1 Rock anchors, 1130-2 Rock cuts, 640-4 geotechnical investigation, 510-10 Rock walls, 1130-2 Roundabouts, 915-1 access, 915-14 access control, modified, 1430-9 access control, partial, 1430-6 bicycle...	{ "page_id": null, "source": 7320, "title": "from dpo" }
Design Manual M 22-01 Index May 2006 Page 1 service walkways, 820-4 structure mounted sign mountings, 820-3 vertical clearance, 820-4 Signing bridge clearance, 1120-3 Silt fence, 530-6 Single point (urban) interchange, 940-3 Single slope barrier, 710-15 Site data for structures, 1110-1 additional data for grade separa...	{ "page_id": null, "source": 7320, "title": "from dpo" }
runoff for ramp curves, 642-4 shared use path, 1020-13 Surface Transportation Program, 120-7 Index Design Manual M 22-01 Page 16 May 2006 Surfacing materials, 510-1 investigation, 510-1 pavement structure, 520-1 surfacing report, 510-14 Surveying and mapping during design and development of the PS&E, 1440-1 during the ...	{ "page_id": null, "source": 7320, "title": "from dpo" }
grades, 1060-11 intersection, 1060-12 lane widths, 1060-11 park and ride lots, 1060-1 park and ride stops, 1055-5 passenger amenities, 1060-10 Design Manual M 22-01 Index May 2006 Page 1 paving sections, 1060-11 priority preemption, 850-11 roundabouts, 915-14 stops, 1055-4 transfer/transit centers, 1060-7 vehicle char...	{ "page_id": null, "source": 7320, "title": "from dpo" }
510-8 noise, 1140-2 retaining, 1130-1 Warrants, 850-4, 1010-2 Washington State Patrol median crossovers, 960-2 weigh sites, 1040-1 Index Design Manual M 22-01 Page 1 May 2006 Washingtons Transportation Plan (WTP), 120-4 Waste Sites, R/W, 1410-2 Water-filled barriers, 810-9 Weaving sections, 940-9 Weigh sites, 1040-1 f...	{ "page_id": null, "source": 7320, "title": "from dpo" }
Title: URL Source: Markdown Content: DISTRIBUTED ALGORITHMS 2010/2011 # Exercise Session 7 Terminating Reliable Broadcast November 15, 2010 ## Problem 1 Can we implement TRB with the eventually perfect failure detector P , if we assume that at least one process can crash? See Solution 6.10 in the book. ## Problem 2 S...	{ "page_id": null, "source": 7321, "title": "from dpo" }
Title: URL Source: Markdown Content: # Failure Detection vs Group Membership in Fault-Tolerant Distributed Systems: Hidden Trade-Offs Andr´ e Schiper Ecole Polytechnique F´ ed´ erale de Lausanne (EPFL) 1015 Lausanne, Switzerland andre.schiper@epfl.ch Abstract. Failure detection and group membership are two important c...	{ "page_id": null, "source": 7321, "title": "from dpo" }
in the asynchronous model), A is also correct in a more re-stricted model (e.g. in the synchronous model). Clearly, it is advantageous to develop algorithms for the most general system model. Unfortunately, it has been proven that a very basic fault-tolerant problem, the con-sensus problem, cannot be solved by a determ...	{ "page_id": null, "source": 7321, "title": "from dpo" }
—which requires that each faulty process is eventually suspected forever by each correct process — and (2) by eventual weak accuracy — which requires that eventually there exists some correct process that is no longer suspected by any correct process. The failure detector model has allowed a very important result to be...	{ "page_id": null, "source": 7321, "title": "from dpo" }
which a new is provided to a process is called the install event. A process may leave the group as a result of an explicit leave request	{ "page_id": null, "source": 7321, "title": "from dpo" }
Title: The Weakest Failure Detector for Genuine Atomic Multicast URL Source: Markdown Content: # The Weakest Failure Detector for Genuine Atomic Multicast # Pierre Sutra # Telecom SudParis, Palaiseau, France Abstract Atomic broadcast is a group communication primitive to order messages across a set of distributed proc...	{ "page_id": null, "source": 7321, "title": "from dpo" }
systems organizing principles; General and reference → Reliability Keywords and phrases Failure Detector, State Machine Replication, Consensus Digital Object Identifier 10.4230/LIPIcs.DISC.2022.35 Related Version Extended Version : # 1 Introduction Context. Multicast is a fundamental group communication primitive use...	{ "page_id": null, "source": 7321, "title": "from dpo" }
International Proceedings in Informatics > Schloss Dagstuhl – Leibniz-Zentrum für Informatik, Dagstuhl Publishing, Germany 35:2 The Weakest Failure Detector for Genuine Atomic Multicast > Table 1 About the weakest failure detector for atomic multicast. (√√ = strongly genuine ) > Genuiness Order Weakest > ×Global Ω∧Σ[8,...	{ "page_id": null, "source": 7321, "title": "from dpo" }
question in distributed computing literature [ 18 ]. In particular, the seminal work in [ 8] shows that a leader oracle ( Ω) is the weakest failure detector for consensus when a majority of processes is correct. If any processes might fail, then a quorum failure detector ( Σ) is required in addition to Ω . A failure de...	{ "page_id": null, "source": 7321, "title": "from dpo" }
failure detector that informs each P. Sutra 35:3 process in g ∪ h when g ∩ h is faulty. The strongly genuine variation requires a message to be delivered when its destination group runs in isolation. In that case, the weakest failure detector is at least μ ∧ (∧g,h ∈G Ωg∩h). This value is attained when F = ∅. Outline of...	{ "page_id": null, "source": 7321, "title": "from dpo" }
to a set of processes denoted by dst (m). This set is named the destination group of m. When a process executes deliver (m), it delivers message m, typically to an upper applicative layer. Consider two messages m and m′ and some process p ∈ dst (m) ∩ dst (m′). Relation m p 7→ m′ captures the local delivery order at pro...	{ "page_id": null, "source": 7321, "title": "from dpo" }
(m)}. For some process p, G(p) denotes the destination groups in G that contain p. Two groups g and h are intersecting when g ∩ h̸ = ∅. What can be sent and to who. The process that executes multicast (m) is the sender of m, denoted src (m). As usual, we consider that processes disseminate different messages (i.e., src...	{ "page_id": null, "source": 7321, "title": "from dpo" }
strategy defeats the core purpose of atomic multicast and is thus not satisfying from a performance perspective. To rule out this class of solutions, Guerraoui and Schiper [ 25 ]introduce the notion of genuine atomic multicast. These protocols satisfy the minimality property defined below. (Minimality) In every run R o...	{ "page_id": null, "source": 7321, "title": "from dpo" }
cyclic families f={g1, g 2, g 3}and f′={g1, g 3, g 4}. In what follows, F denotes all the cyclic families in 2G . Given a destination group g, F(g) are the cyclic families in F that contain g. For some process p, F(p) are the cyclic families f such that p belongs to some group intersection in f (that is, ∃g, h ∈ f. p ∈...	{ "page_id": null, "source": 7321, "title": "from dpo" }
below. The quorum failure detector ( Σ) captures the minimal amount of synchrony to implement an atomic register. When a process p queries at time t a detector of this class, it returns a non-empty subset of processes Σ( p, t ) ⊆ P such that: (Intersection) ∀p, q ∈ P . ∀t, t ′ ∈ N. Σ( p, t ) ∩ Σ( q, t ′)̸ = ∅ (Liveness...	{ "page_id": null, "source": 7321, "title": "from dpo" }
2 2 35:6 The Weakest Failure Detector for Genuine Atomic Multicast Accuracy ensures that if some cyclic family f is not output at p and p belongs to it, then f is faulty at that time. Completeness requires that eventually γ does not output forever a faulty family at the correct processes that are part of it. Hereafter,...	{ "page_id": null, "source": 7321, "title": "from dpo" }
t) = F (t) ∩ P . Then, DP (F ) equals D(F ∩ P ) at p ∈ P , and the mapping p × N → ⊥ elsewhere. To illustrate this definition, Ω{p} is the trivial failure detector that returns p at process p. Another example is given by Σ{p1,p 2} which behaves as Σ over P = {p1, p 2}. The candidate. Our candidate failure detector is μ...	{ "page_id": null, "source": 7321, "title": "from dpo" }
m′. This variation requires that if m and m′ are multicast to the same group, then m ≺ m′, or the converse, holds. Proposition 1 below establishes that this variation is as difficult as (vanilla) atomic multicast. Building upon this insight, this section depicts a solution to group sequential atomic multicast using fai...	{ "page_id": null, "source": 7321, "title": "from dpo" }
our knowledge, this is only algorithm with [ 34 ] that tolerates arbitrary failures. Algorithm 1 is composed of a set of actions. An action is executable once its preconditions ( pre: ) are true. The effects ( eff: ) of an action are applied sequentially until it returns. Algorithm 1 uses a log per group and per group ...	{ "page_id": null, "source": 7321, "title": "from dpo" }
or they both occupy the same slot and d < d ′, for some a priori total order ( <) over the data items. Variables. Algorithm 1 employs two types of shared objects at a process. First, for any two groups h and h′ to which the local process belongs, Algorithm 1 uses a log LOG h∩h′ (line 2). Notice that, when h = h′, the l...	{ "page_id": null, "source": 7321, "title": "from dpo" }
shared objects. D I S C 2 0 2 2 35:10 The Weakest Failure Detector for Genuine Atomic Multicast Algorithm 2 Emulating Σ∩g∈Gg – code at process p. > 1: variables: > 2: (Ag,x )g∈G,x ⊆g.p ∈x > 3: (Qg )g∈G ← λg. {g} > 4: (qr g )g∈G ← λg.g > 5: for all g ∈ G, x ⊆ g : p ∈ x do > 6: let m such that dst (m) = g ∧ payload (m) =...	{ "page_id": null, "source": 7321, "title": "from dpo" }
directly from the communication primitive. Algorithm 2 presents such a construction. This algorithm can be seen as an extension of the work of Bonnet and Raynal [ 6] to extract Σk when k-set agreement is solvable. Algorithm 2 emulates Σ∩g∈Gg , where G ⊆ G is a set of at most two intersecting destination groups. At a pr...	{ "page_id": null, "source": 7321, "title": "from dpo" }
(line 14). As stated in Theorem 2 below, these quorums must intersect at any two processes in ∩g∈Gg. ▶ Theorem 2. Algorithm 2 implements Σ∩g∈Gg in E. # 5.2 Emulating γ Target systems. A process p is failure-prone in environment E when for some failure pattern F ∈ E, p ∈ Faulty (F ). By extension, we say that P ⊆ P is f...	{ "page_id": null, "source": 7321, "title": "from dpo" }
failure detector γ in Algorithm 3. For each closed path π ∈ cpaths (f) with π ∩ π failure-prone in E, Algorithm 3 maintains two variables: an instance Aπ of the multicast algorithm A, and a flag failed [π]. Variable Aπ is used to detect when a group intersection visited by π is faulty. It this happens, the flag failed ...	{ "page_id": null, "source": 7321, "title": "from dpo" }
variables: > 2: (Aπ)π// ∀f∈ F (p).∀π∈cpaths (f). p / ∈π ∩π[\|π\| − 2] > 3: failed [π]←λπ. false > 4: for all Aπ:p∈π ∩π do > 5: Aπ.multicast (p, 0) to π > 6: signal (π, i ):= > 7: pre: Aπ.deliver (−, i ) > 8: i 9: eff: send (π, i )to f > 10: Aπ.multicast (p, i + 1) to π[i+ 1] > 11: update (π):= > 12: pre: ∃π′≡π. rcv (π, ...	{ "page_id": null, "source": 7321, "title": "from dpo" }
and each replica maintains its own local copy of the machine. Commands accessing the service are funneled through the ordering primitive before being applied at each replica on the local copy. SMR protocols must satisfy linearizability [ 28 ]. However, as observed in [ 3 ], the common definition of atomic multicast is ...	{ "page_id": null, "source": 7321, "title": "from dpo" }
P are faulty or not. In detail, this failure detector returns a boolean which ensures that: (Accuracy) ∀p ∈ P . ∀t ∈ N. 1P (p, t ) ⇒ P ⊆ F (t) (Completeness) ∀p ∈ Correct . ∀t ∈ N. P ⊆ F (t) ⇒ ∃ τ ∈ N. ∀t′ ≥ τ. 1P (p, t ′) For simplicity, we write 1g∩h the indicator failure detector restricted to the processes in g ∪ h...	{ "page_id": null, "source": 7321, "title": "from dpo" }
information than γ regarding the correctness of g ∩ h. As a consequence, the modified algorithm solves (group sequential) atomic multicast. D I S C 2 0 2 2 35:14 The Weakest Failure Detector for Genuine Atomic Multicast Now, to see why such a solution is strict, consider two messages m and m′ that are delivered in a ru...	{ "page_id": null, "source": 7321, "title": "from dpo" }
of the destination group, creating a delay that harms performance and reduces parallelism [ 17 , 1]. In this section, we explore a stronger form of genuineness, where groups are able to deliver messages independently. We prove that, similarly to the strict variation, this requirement demands more synchrony than μ from ...	{ "page_id": null, "source": 7321, "title": "from dpo" }
appears in . Each process constructs a directed acyclic graph G by sampling the failure detector D and exchanging these samples with other processes. A path π in G induces multiple runs of A that each process locally simulates. A run starts from some initial configuration. In our context, the configurations I = {I1, . ...	{ "page_id": null, "source": 7321, "title": "from dpo" }
be correct. Otherwise, there exists a bivalent root of some tree Υi such that for g (respectively, h) a correct process multicasts a message to g (resp., h) in Ii. In this case, similarly to [ 8], there exists a decision gadget in the simulation tree Υi. This gadget is a sub-tree of the form (S, S ′, S ′′ ), with S biv...	{ "page_id": null, "source": 7321, "title": "from dpo" }
(at least) two processes. In detail, the class Uk are all the failure detectors D that are k-unreliable, that is they cannot distinguish any pair of failure patterns F and F ′, as long as the faulty processes in F and F ′ are members of a subset W of size k (the “wrong” subset). The result in [ 25 ] is a corner case of...	{ "page_id": null, "source": 7321, "title": "from dpo" }
certain cases, e.g., if two groups intersect on a single process p, then p must be reliable. In Figure 1, this happens with process p2. In contrast, to these prior solutions, Algorithm 1 tolerates any number of failures. This is also the case of which relies on a perfect failure detector. Regarding strongly genuine at...	{ "page_id": null, "source": 7321, "title": "from dpo" }
and Networks, DSN 2017, Denver, CO, USA, June 26-29, 2017 , pages 37–48. IEEE Computer Society, 2017. doi:10.1109/DSN.2017.15 . 11 James A. Cowling and Barbara Liskov. Granola: Low-overhead distributed transaction coordination. In Gernot Heiser and Wilson C. Hsieh, editors, 2012 USENIX Annual Technical Conference, Bost...	{ "page_id": null, "source": 7321, "title": "from dpo" }
EuroSys ’21: Sixteenth European Conference on Computer Systems, Online Event, United Kingdom, April 26-28, 2021 , pages 178–193. ACM, 2021. doi:10.1145/3447786. 3456236 . 18 Felix C. Freiling, Rachid Guerraoui, and Petr Kuznetsov. The failure detector abstraction. ACM Comput. Surv. , 43(2), February 2011. doi:10.1145/1...	{ "page_id": null, "source": 7321, "title": "from dpo" }
1994. 27 Maurice Herlihy. Wait-free synchronization. ACM Transactions on Programming Languages and Systems , 11(1):124–149, January 1991. doi:10.1145/114005.102808 . 28 Maurice Herlihy and Jeannette Wing. Linearizability: a correcteness condition for concurrent objects. ACM Transactions on Programming Languages and Sys...	{ "page_id": null, "source": 7321, "title": "from dpo" }
Notes in Computer Science , pages 481–495. Springer, 2008. doi:10.1007/978-3-540-92221-6_30 .P. Sutra 35:19 35 Nicolas Schiper, Pierre Sutra, and Fernando Pedone. Genuine versus non-genuine atomic multicast protocols for wide area networks: An empirical study. In 28th IEEE Symposium on Reliable Distributed Systems (SRD...	{ "page_id": null, "source": 7321, "title": "from dpo" }
Title: Tame the Wild with Byzantine Linearizability: Reliable Broadcast, Snapshots, and Asset Transfer URL Source: Markdown Content: # Tame the Wild with Byzantine Linearizability: Reliable Broadcast, Snapshots, and Asset Transfer # Shir Cohen # Technion, Haifa, Israel # Idit Keidar # Technion, Haifa, Israel Abstract ...	{ "page_id": null, "source": 7322, "title": "from dpo" }
implement reliable read-write shared memory registers via message passing even if a fraction of the servers are Byzantine [ 1, 21 , 24 , 19 ]. As a result, as long as the client processes using the service are not malicious, any fault-tolerant object that can be constructed using registers can also be implemented in th...	{ "page_id": null, "source": 7322, "title": "from dpo" }
cannot be implemented without a majority of correct processes constantly taking steps. Asset transfer has wait-free implementations from both reliable broadcast [ 7] and snapshots 17 and thus the same lower bound applies to reliable broadcast and snapshots as well. In Section 6, we present a Byzantine linearizable re...	{ "page_id": null, "source": 7322, "title": "from dpo" }
than servers, particularly in the so-called permissioned model where servers are trusted and clients are ephemeral. In summary, we make the following contributions: Formalizing Byzantine linearizability for any object with a sequential specification. Proving that some of the most useful building blocks in distributed c...	{ "page_id": null, "source": 7322, "title": "from dpo" }
and Byzantine snapshot, (from which Byzantine lattice agreement can be constructed [ 6 ]), with resilience n = 2 f + 1 .The asset transfer object we discuss in this paper was introduced by Guerraoui et al. [ 17 , 15 ]. Their work provides a formalization of the cryptocurrency definition [ 23 ]. The highlight of their w...	{ "page_id": null, "source": 7322, "title": "from dpo" }
balance is always at least 10, and performs a transfer with amount 5 after another process deposits 5 into its account, increasing its balance to 15. Using the protocol in [ 7], another process might observe i’s balance as 5 if it sees i’s outgoing transfer before the causally preceding deposit. Because our solution is...	{ "page_id": null, "source": 7322, "title": "from dpo" }
process is equipped with a public-private key pair used to sign data and verify signatures of other processes. We denote a value v signed by process i as ⟨v⟩i. Executions and Histories. We discuss algorithms emulating some object O from lower level objects (e.g., registers). An algorithm is organized as methods of O. A...	{ "page_id": null, "source": 7322, "title": "from dpo" }
(1) after removing some pending operations from H and completing others by adding matching responses, it contains the same invocations and responses as H′, (2) H′ preserves the partial order ≺H , and (3) H′ satisfies o’s sequential specification. f-resilient. An algorithm is f-resilient if as long as at most f processe...	{ "page_id": null, "source": 7322, "title": "from dpo" }
in a way that provides consistency to correct processes. We linearize the correct processes’ operations and offer a degree of freedom to embed additional operations by Byzantine processes. We denote by H\|correct the projection of a history H to all correct processes. We say that a history H is Byzantine linearizable if...	{ "page_id": null, "source": 7322, "title": "from dpo" }
object. The reason is that the algorithm relies on correct processes being able to perform “double-collect” meaning that at some point a correct process manages to read all registers twice without witnessing any changes. While this is true in the crash-failure model, in the Byzantine model this is not the case as the a...	{ "page_id": null, "source": 7322, "title": "from dpo" }
intrusions, it is less appropriate for settings like cryptocurrencies, where Byzantine behavior cannot be expected to eventually stop. # 5 Lower Bound on Resilience In shared memory, one typically aims for wait-free objects, which tolerate any number of process failures. Indeed, many useful objects have wait-free imple...	{ "page_id": null, "source": 7322, "title": "from dpo" }
shown in Figure 1. Let σ1 be an execution where, only processes in A ∪ { p1} take steps. First, z performs transfer( z, p 1,1) . Since up to f processes may be faulty, the operation completes, and by the object’s sequential specification, it is successful (returns true). Then, p1 performs transfer( p1, q 1,1) . By f -r...	{ "page_id": null, "source": 7322, "title": "from dpo" }
all processes in A ∪ { p2} are Byzantine by executing σ2, having all processes in A ∪ { p2} reset their memory, executing σ1, and then having z and p2 restore their registers to their state at the end of σ2. At this point, the state of z and p2 is the same as it was at the end of σ2, the state of processes in A \ { z} ...	{ "page_id": null, "source": 7322, "title": "from dpo" }
q1, whereas σ4 the opposite is true. This is a contradiction. ◀ Guerraoui et al. [ 17 ] use an atomic snapshot to implement an asset transfer object in the crash-fault shared memory model. In addition, they handle Byzantine processes in the message passing model by taking advantage of reliable broadcast. In Appendix A ...	{ "page_id": null, "source": 7322, "title": "from dpo" }
f -resilient Byzantine linearizable implementation of asset transfer in a system with n ≥ 2f + 1 processes, f of which can be Byzantine, using only SWMR registers if less than f + 1 correct processes take steps infinitely often. Proof. Assume by way of contradiction that there exists an f -resilient Byzantine lineariza...	{ "page_id": null, "source": 7322, "title": "from dpo" }
given for message passing systems [ 10 ], requires the following properties: Validity: If a correct process i broadcasts (ts, m ) then all correct processes eventually deliver m from process i in timestamp ts .Agreement: If a correct process delivers m from process i in timestamp ts , then all correct processes eventua...	{ "page_id": null, "source": 7322, "title": "from dpo" }
messages are subsequently pulled then according to Definition 6 all correct processes deliver m, providing validity. For agreement, if a correct process invokes deliver(j,ts) that returns m and all messages are later pulled by all correct processes, it follows that all correct processes also invoke deliver(j,ts) and ev...	{ "page_id": null, "source": 7322, "title": "from dpo" }
at first a process reads the last value written to a send register (line 16). If the value is a signed pair of a message and a timestamp, refresh then copies it to the process’s echo register in line 18. In the echo register, the value remains as evidence, preventing conflicting values (sent by Byzantine processes) fro...	{ "page_id": null, "source": 7322, "title": "from dpo" }
(j,ts) > 9: refresh() > 10: if ∃k ∈ Π and v s.t. ⟨⟨ ts, v ⟩j , σ ⟩ ∈ deliver k where σ is a set of f + 1 signatures on ⟨ready, ⟨ts, v ⟩j ⟩ then > 11: deliver i ← deliver i ∪ {⟨⟨ ts, v ⟩j , σ ⟩} > 12: return v > 13: return ⊥ > 14: procedure refresh > 15: for j ∈ [n] do > 16: m ← send j > 17: if ∄ts , val s.t. m = ⟨ts, v...	{ "page_id": null, "source": 7322, "title": "from dpo" }
utilize a reliable broadcast primitive to construct a Byzantine snapshot object with resilience n > 2f . # 7.1 Snapshot Object A snapshot [ 2 ] is represented as an array of n shared single-writer variables that can be accessed with two operations: update(v) , called by process i, updates the ith entry in the array and...	{ "page_id": null, "source": 7322, "title": "from dpo" }
as the supremum of the collects in their start messages. We achieve optimal resilience by waiting for only f + 1 processes to send the same set. Although there is not necessarily a correct process in the intersection of two sets of size f + 1 ,we leverage the fact that reliable broadcast prevents equivocation to ensure...	{ "page_id": null, "source": 7322, "title": "from dpo" }
(v) > 2: for j ∈ [n] do ▷ collect current memory state > 3: update-collect( collected j ) > 4: ts i ← ts i + 1 > 5: collected i[i] ← ⟨ ts i, v ⟩i ▷ update local component of collected > 6: procedure snapshot > 7: for j ∈ [n] do ▷ collect current memory state > 8: update-collect( collected j ) > 9: c ← collected i > 10:...	{ "page_id": null, "source": 7322, "title": "from dpo" }
31) and broadcasts it to all processes at line 33. S. Cohen and I. Keidar 18:13 Algorithm 3 Byzantine Snapshot auxiliary procedures: code for process i. 19: procedure minimum-saved (auxnum) 20: S ← { s\|∃ j ∈ [n], s = savesnap j [auxnum ].snap and savesnap j [auxnum ].proof is a valid proof of s} 21: if S = ∅ then 22: r...	{ "page_id": null, "source": 7322, "title": "from dpo" }
i[p] − 1] 49: rts i[p] ← rts i[p] + 1 50: if received f + 1 round-r messages for the first time then 51: r ← r + 1 52: broadcast( r, ⟨senders i⟩i) 53: if ∃s s.t. \|{ j\| seen i[j][ s] = senders i}\| = f + 1 then ▷ stability condition 54: r ← 0 55: senders i← ∅ 56: ∀j ∈ [n], k ∈ N, seen i[j][k] ← ∅ 57: cached ← minimum-sav...	{ "page_id": null, "source": 7322, "title": "from dpo" }
start messages (line 52). Process i maintains a local map seen [j][ r] that maps a process j and a round r to the set of processes that j reported to have received start messages from in rounds 1–r (line 48), but only if i has received start messages from all the reported processes (line 45). By doing so, we ensure tha...	{ "page_id": null, "source": 7322, "title": "from dpo" }
case some correct processes complete a snapshot-aux instance before all do, we add a helping mechanism. Whenever a correct process successfully completes snapshot-aux, it stores its result in a savesnap map, with the auxnum as the key (either at line 24 or at line 59). This way, once one correct process returns from sn...	{ "page_id": null, "source": 7322, "title": "from dpo" }
has timestamp ts if s[j].ts ≥ ts . If multiple updates are linearized to the same point (before the same snapshot), we order them by their start times. Finally, we add updates by Byzantine processes as follows: We add update(v) by a Byzantine process j if there is a linearized snapshot that returns s and s[j].val = v. ...	{ "page_id": null, "source": 7322, "title": "from dpo" }
is, if a snapshot operation sees the second update, it sees the first one. ▶ Lemma 13. If update1 by process i precedes update2 by process j and a snapshot operation snap by a correct process sees update2, then snap sees update1 as well. Finally, the next lemma proves the liveness of our algorithm. ▶ Lemma 14. (Livenes...	{ "page_id": null, "source": 7322, "title": "from dpo" }
asset transfer. Thus, we proved a tight bound on the resilience of emulations of asset transfer, snapshot, and reliable broadcast. Our paper deals with feasibility results and does not focus on complexity measures. In particular, we assume unbounded storage in our constructions. We leave the subject of efficiency as an...	{ "page_id": null, "source": 7322, "title": "from dpo" }
and secure distributed programming . Springer Science & Business Media, 2011. 11 Miguel Castro, Barbara Liskov, et al. A correctness proof for a practical byzantine-fault-tolerant replication algorithm. Technical report, Technical Memo MIT/LCS/TM-590, MIT Laboratory for Computer Science, 1999. 12 Miguel Castro, Barbara...	{ "page_id": null, "source": 7322, "title": "from dpo" }
Alvisi, and Michael Dahlin. Minimal byzantine storage. In International Symposium on Distributed Computing , pages 311–325. Springer, 2002. 22 Achour Mostéfaoui, Matoula Petrolia, Michel Raynal, and Claude Jard. Atomic read/write memory in signature-free byzantine asynchronous message-passing systems. Theory of Com-put...	{ "page_id": null, "source": 7322, "title": "from dpo" }
the snapshot taken to compute the balance for each transaction. D I S C 2 0 2 1 18:18 Tame the Wild with Byzantine Linearizability > Algorithm 4 Byzantine Asset Transfer: code for process i. shared Byzantine snapshot: S initial– immutable array of initial balances local variables: txns i – sets of outgoing transaction,...	{ "page_id": null, "source": 7322, "title": "from dpo" }
Title: The consensus number of a cryptocurrency (extended version) URL Source: Markdown Content: > Distributed Computing # The consensus number of a cryptocurrency (extended version) Rachid Guerraoui 1 · Petr Kuznetsov 2 · Matteo Monti 1 · Matej Pavlovic 1 · Dragos-Adrian Seredinschi 1 > Received: 10 February 2020 / ...	{ "page_id": null, "source": 7322, "title": "from dpo" }
results are applicable to both the permissioned (private) and permissionless (public) setting, as normally their differentiation is hidden by the abstractions on top of which our algorithms are based. Keywords Distributed computing · Cryptocurrency · Consensus · Blockchain · Decentralized payments # 1 Introduction The ...	{ "page_id": null, "source": 7322, "title": "from dpo" }
energy-efficiency [5,8,25,47]. Not surprisingly, the consensus module is a major bottleneck in blockchain-based protocols [28,43,47]. A close look at Nakamoto’s original paper reveals that the central issue in implementing a decentralized asset transfer system (i.e., a cryptocurrency) is preventing double-spending , i....	{ "page_id": null, "source": 7322, "title": "from dpo" }
with-drawal from an account is validated by relating the withdrawn amount with the incoming transfers found in the memory snapshot. Intuitively, as at most one withdrawal can be active on a given account at a time (as the account’s owner is a single sequential process), it is safe to declare the validated opera-tion as...	{ "page_id": null, "source": 7322, "title": "from dpo" }
relies on a secure broadcast primitive that ensures uniform reliable delivery with only weak ordering guarantees [36,37], circumventing hurdles imposed by con-sensus. In the k-shared case, our results imply that to execute some form of smart contract involving k users, consensus is only needed among these k nodes and n...	{ "page_id": null, "source": 7322, "title": "from dpo" }
a set of states, q0 ∈ Q is an initial state, O is a set of operations, R is a set of responses and Δ ⊆ Q × Π × O × Q × R is a relation that associates a state, a process identifier and an operation to a set of possible new states and corresponding responses. We assume that Δ is total on the first three elements. A hist...	{ "page_id": null, "source": 7322, "title": "from dpo" }
i.e., a sequence of distinct invocations and responses, labelled with process identifiers and unique sequence numbers. A projection of a history H to process p, denoted H \| p is the subsequence of elements of H labelled with p. An invocation o by a process p is incomplete in H if it is not followed by a response in H \|...	{ "page_id": null, "source": 7322, "title": "from dpo" }
asset-transfer object type associated with A and μ as a tuple (Q, q0, O, R, Δ) , where: – The set of states Q is the set of all possible maps q : A → N. Intuitively, each state of the object assigns each account its balance .– The initialization map q0 : A → N assigns the initial balance to each account. – Operations a...	{ "page_id": null, "source": 7322, "title": "from dpo" }
( p /∈ μ( a) ∨ q(a) < x) ∧ q′ = q ∧ r = false ;A transfer fails (having false as response) without modifying the state ( q′ = q) if the invoking process is not the owner of account a or if a has insufficient balance. – o = read (a) ∧ q = q′ ∧ r = q(a).Operation read (a) simply returns the balance of a and leaves the ac...	{ "page_id": null, "source": 7322, "title": "from dpo" }
of read-write registers in a shared memory system with crash failures) , it follows that the asset-transfer type also has consensus number 1. Consider an asset-transfer object associated with a set of accounts A and an ownership map μ where ∀a ∈ A, \|μ( a)\| ≤ 1. Our implementation is described in Fig. 1. Every process p...	{ "page_id": null, "source": 7322, "title": "from dpo" }
operation and return true . Otherwise, the operation returns false . Theorem 1 The asset-transfer object type has a wait-free implementation in the read-write shared memory model. Proof Fix an execution E of the algorithm in Fig. 1. Atomic snapshots can be wait-free implemented in the read-write shared memory model . A...	{ "page_id": null, "source": 7322, "title": "from dpo" }
of L. Let L be the legal prefix of the first invocations and op be the ( + 1)st operation of L. Let op be invoked by process p. The following cases are possible: – op is a read (a): the snapshot taken at the linearization point of op contains all successful transfers concerning a in L. By the induction hypothesis,...	{ "page_id": null, "source": 7322, "title": "from dpo" }
is 2 Analogously to balance (a, S) that computes the balance for account > abased on the transfers contained in snapshot S,balance (a,L), if Lis a sequence of operations, computes the balance of account abased on all transfers in L. > Fig. 2 Wait-free implementation of consensus among kprocesses using ak-shared asset-t...	{ "page_id": null, "source": 7322, "title": "from dpo" }
input in a register (line 1). Process p then tries to perform a transfer from account a to another account. The amount withdrawn this way from account a is chosen specifically such that: – only one transfer operation can ever succeed, and – if the transfer succeeds, the remaining balance on a will uniquely identify pro...	{ "page_id": null, "source": 7322, "title": "from dpo" }
k-shared asset-transfer to k-consensus. A k-consensus object exports a single opera-tion propose that, the first k times it is invoked, returns the argument of the first invocation. All subsequent invocations return ⊥. Given that k-consensus is known to have consensus number exactly k , a wait-free algorithm implementi...	{ "page_id": null, "source": 7322, "title": "from dpo" }
objects. Code for process p To read the balance of account a, a process takes a snapshot of AS , and then sums the initial balance q0(a) and amounts of all successful incoming transfers, and subtracts the amounts of successful outgoing transfers found in AS . We say that a successful transfer t x is in a snapshot AS (d...	{ "page_id": null, "source": 7322, "title": "from dpo" }
indefinitely helping other processes, given only a finite number of possible older operations pro-posed by other processes. Then, p takes a snapshot of AS and checks whether account a has sufficient balance according to the state rep-resented by the snapshot, and equips the transfer with a corresponding success / failu...	{ "page_id": null, "source": 7322, "title": "from dpo" }
atomic snapshot and k-consensus operation has been lin-earized). Intuitively, we include in Ops all operations that took effect , either by returning a response to the user or by affecting other operations. Recall that every such transfer operation was agreed upon in an instance of k-consensus, let it be kC o. Therefor...	{ "page_id": null, "source": 7322, "title": "from dpo" }

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.