| • Kelsey Hightower's background and experience with Google Cloud technologies, Golang, and Kubernetes
|
| • The intersection of sysadmin skills and coding, and how Kelsey brings his operational expertise to development roles
|
| • Kubernetes as a system that codifies ops people's expertise, requiring operational knowledge to deploy and maintain
|
| • Abstraction in Kubernetes: abstracting users from cluster operation concerns, but still requiring focus on retry logic, logging, etc.
|
| • Kelsey's analogy of Kubernetes as the runtime for infrastructure, much like Go's runtime for applications
|
| • The comparison of Kubernetes to a kernel or operating system, with potential implications for self-deploying applications and system call interfaces
|
| • The relationship between Golang and Kubernetes: how Go's strengths (cross-platform executables, concurrency) make Kubernetes special
|
| • Discussing the prevalence of Go-written tools in the Kubernetes ecosystem
|
| • Explaining Kubernetes' plugin architecture and how it allows for custom extensions without recompiling the system
|
| • Describing the concepts of desired state, reconciliation, and controllers/schedulers in Kubernetes
|
| • Comparing Kubernetes to other systems like Puppet, CFEngine, Chef, and Ansible in terms of declarative vs. imperative programming models
|
| • Discussing the use of labels and soft/hard requirements for scheduling in Kubernetes 1.4
|
| • Kubernetes as a framework for building distributed systems
|
| • Kubernetes is not just about deploying containers, but can be used to build new systems and workflows
|
| • Core object types in Kubernetes (node, pod, controller, service) and how they enable extensions such as deployments and replica sets
|
| • Extending the system with user-defined types at runtime using third-party resources
|
| • Importance of abstraction layers on top of core objects to facilitate tasks and interactions with the cluster
|
| • Exploring and integrating Kubernetes into existing applications and workflows for developers who may not be familiar with DevOps tools
|
| • Docker vs Kubernetes: Developers may be less interested in cluster management and deployment, focusing on app development
|
| • Kubernetes concepts: Automated scheduling, resource allocation, scaling, and load balancing
|
| • Twelve-Factor App principles: Importance of decoupling from machine dependencies, using environment variables, logs to STDOUT
|
| • Integrating Twelve-Factor with Kubernetes: Using config maps and secrets for configuration injection
|
| • Containerization vs cluster management: Understanding the differences between Docker, Kubernetes, and DevOps practices
|
| • Kubernetes can be overkill for simple applications and comes with its own set of management tasks
|
| • Kelsey Hightower notes that even if a well-designed application doesn't need Kubernetes initially, it's becoming an expected feature as customers demand high uptime and availability
|
| • Google Cloud's GKE (Google Kubernetes Engine) is mentioned as a commercial offering of Kubernetes and provides deep integrations with the Google Cloud platform
|
| • The Pokémon GO example is cited as a real-world success story for Kubernetes in handling high traffic and revenue-generating workloads
|
| • Kelsey Hightower discusses his idea of self-deploying Go applications, where an app can automatically deploy itself on Kubernetes without manual YAML files or complex configuration
|
| • He showcases his prototype "Hello, Universe" which demonstrates this concept by allowing users to scale and manage their application across multiple clusters with ease.
|
| • Kubernetes allows users to submit a configuration for a channel lineup or cable television and have the system manage it without worrying about hardware or software details
|
| • The system provides concurrency capabilities "for free" in terms of not requiring additional setup or infrastructure
|
| • Cluster Federation is a new feature that enables managing multiple clusters as a single cluster, making it easier to handle large, complex deployments
|
| • Live demos can be an effective way for presenters to demonstrate concepts and inspire others to try them themselves
|
| • Kelsey's ability to explain complex topics in an engaging way
|
| • Support and development of the Go language for projects like Kubernetes
|
| • Native vendoring directory and third-party dependency management
|
| • Importance of community contributions, including Go experts and SIGs
|
| • Kelsey's pragmatic approach to working with Go's limitations
|
| • Compilation of best practices from the Kubernetes project
|
| • Examples of good engineering practices, such as end-to-end testing
|
| • The Go Build Template, a make file and Go project structure extracted from Kubernetes
|
| • Discussing the importance of API versioning and migration from alpha to stable
|
| • Mentioning Kubernetes' excellent testing practices for distributed applications
|
| • Debate on idiomatic use of languages in large projects and the value of documenting best practices
|
| • Shoutouts to free software maintainers, including Dave Cheney's packages (github.com/pkg/errors and github.com/felixge/pidctrl)
|
| • Recommendation of a resource: Golang Spec compilation of blog posts for learning Go
|
| • Scaling a Go project to use all available resources
|
| • Importance of documentation in software development
|
| • Shoutouts to Ben Johnson and BoltDB for excellent documentation
|
| • The value of example code and projects in learning new concepts
|
| • Wrap-up and show sponsor thank-yous |
