Kubernetes - Changing the way we build platforms

Presented at docker meetup in Melbourne 23/09/2015

Transcript

1. Kubernetes Changing the way we build platforms
2. Who is this guy? @wolfeidau on twitter github.com/wolfeidau www.wolfe.id.au DevOps @ http://www.versent.com.au/
3. Greek for “Helmsman”; also the root of the word “Governor”
4. Kubernetes? • Inspired by an internal project at Google, called Borg • Contributors from Google, Mesos and more • 1.0 release a month or so ago
5. Containers • Unix processes not lightweight Virtual Machines • Application + Dependencies = image • Runtime environment (cgroups, namespaces, environment variables)
6. • Build applications in a dedicated build container or CI • Ship build artifacts, not build environments • Rails app container with build tools ~900MB • Rails app container without build tools ~300MB • Go service without build tools ~15MB • Cattle not Pets! Containers
7. • Container management, scheduling, and service discovery. • API driven application management • Agents monitor endpoints for state changes (real-time) • Controllers enforce desired state • Resources (nodes, applications, services) Kubernetes
8. Kubernetes • High level concepts • node • pod • scheduler • replication • service
9. Node • Runs containers and proxies service requests. • docker • kubelet • proxy
10. Pod • Represents a logical application. • One or more containers • Shared namespaces
11. Scheduler • Schedules pods to run on nodes. • Global scheduler for long running jobs • Best fit chosen based on pod requirements • Pluggable
12. Replication Controller • Manages a replicated set of pods. • Creates pods from a template • Ensures desired number of pods are running • Online resizing
13. Replication Controller • Manages a replicated set of pods. • Creates pods from a template • Ensures desired number of pods are running • Online resizing
14. Service • Service discovery for pods. • Proxy runs on each node • Virtual IP per service (avoid port collisions) • Basic round-robin algorithm • Dynamic backends based on label queries
15. Example Pod apiVersion: v1 kind: Pod metadata: labels: db: rethinkdb role: admin name: rethinkdb-admin spec: containers: - image: gcr.io/google_containers/rethinkdb:1.16.0_1 name: rethinkdb env: - name: POD_NAMESPACE valueFrom: fieldRef: fieldPath: metadata.namespace ports: - containerPort: 8080 name: admin-port volumeMounts: - mountPath: /data/rethinkdb_data name: rethinkdb-storage volumes: - name: rethinkdb-storage emptyDir: {}
16. Example Service apiVersion: v1 kind: Service metadata: labels: db: rethinkdb name: rethinkdb-admin spec: ports: - port: 8080 targetPort: 8080 type: LoadBalancer selector: db: rethinkdb role: admin
17. Deployment • Continuous Integration / Deployment • Jenkins • Buildkite • Bamboo
18. Transparency • Logging • ELK (Elastic Search, Logstash, Kibana) • Metrics • InfluxDB and Grafana • Tracing • Twitter Zipkin
19. Resilience • Continuous Failure Testing • Chaos Monkey • Load Testing • Gattling • boom
20. Cloud Providers • Kubernetes Supports • Google Cloud Compute and Container • AWS, being used by Samsung and Others • Openstack • Azure
21. Questions?
22. Links • Kubernetes Site • AWS Setup • CoreOS Kubernetes Vagrant • Kubernetes Rails Deployment • http://fabric8.io/