Share
Embed
In a World of Ephemeral Containers, How Do We Keep Track of Things? @ OpenStack Summit Tokyo
Ian Lewis
October 29, 2015
Transcript
43 slides
-
In a World of Ephemeral Containers, How Do We Keep
Track of Things? Ian Lewis, Developer Advocate, Google Cloud Platform -
Clusters of ephemeral containers Keeping track of things (state!) Concrete
example Secret bonus? 1 2 Agenda 3 4 -
Isolation: Keep jobs from interfering with each other Scheduling: Where
should my job be run? Lifecycle: Keep my job running Discovery: Where is my job now? Constituency: Who is part of my job? Scale-up: Making my jobs bigger or smaller Auth{n,z}: Who can do things to my job? Monitoring: What’s happening with my job? Health: How is my job feeling? Now that we have containers... -
Greek for “Helmsman”; also the root of the word “Governor”
• Container orchestrator • Runs Docker containers • Supports multiple cloud and bare- metal environments • Inspired and informed by Google’s experiences and internal systems • Open source, written in Go Manage applications, not machines Kubernetes -
A small group of tightly coupled containers Example: static site
generator & web server A loop that drives current state towards desired state Example: replication controller A set of running pods that work together Example: load- balanced backends Arbitrary metadata to organize components Example: phase=production role=frontend Pod Replication Controller Service Labels -
• An image is a stack of Read-Only file system
layers. • Usual process: ◦ build ◦ push to repository ◦ pull to execution host ◦ start container from image Images Debian App PHP & Apache Libs -
Implemented on top of a number of (unrelated) Linux APIs:
cgroups: Restrict resources a process can consume CPU, memory, disk IO, ... namespaces: Change a process’s view of the system Network interfaces, PIDs, users, mounts, ... capabilities: Limits what a user can do mount, kill, chown, etc They're processes! Containers -
Containers Read / Write Read / Write Debian App PHP
& Apache Libs Read / Write • A container is a process ◦ started with kernel restrictions ◦ a stack of shared Read-Only file system layers ◦ plus a process specific Read- Write layer • Every new container gets a new Read-Write layer. All containers from the same image start from exactly the same state! -
• It's possible to mount host directories into a container's
filesystem. • These are mutable and do outlive the container. • They're only available on that host. Mounting host directories Debian App PHP & Apache Libs Read / Write host dir -
in a cluster of ephemeral containers application state must exist
outside of the container -
• Networked storage for the cluster • Exposed as a
file system • These are mutable and do outlive the container. • Independent of host node! • Plugins: GCEPersistentDisk, AWSElasticBlockStore, NFS, iSCSI, RBD (Ceph Block Device), Glusterfs Kubernetes PersistentVolumes Debian App PHP & Apache Libs Read / Write -
Run software which stores state outside of the cluster. Manage
it however you like. Connect to it over the network. ex: MySQL managed by DBAs or managed cloud services Patterns Most software expects access to filesystem state across process invocations. Use cluster tools to provide that across the whole cluster. ex: run a MySQL on a filesystem provided by the cluster outside the cluster adapt to run in the cluster cluster native Use software designed for running in clusters - where nodes come and go. Distributed systems. ex: run Cassandra or Riak inside the cluster -
Outside the cluster - run MySQL the usual ways Manage
it yourself • Depends on skills and availability of your admins • May already exist! Subscribe to a managed MySQL service • Automation of backups, replication failover, etc • Limited flexibility -
Adapt to run in the cluster - MySQL in the
cluster Volume • MySQL running in the cluster with data stored in a PersistentVolume. • Data files available regardless of host node • Can restart process on any node (with a short downtime) • Test your workload! -
Cluster native - MySQL on Vitess • Open source MySQL
scaling solution from YouTube • replication, sharding, caching and more • Designed for a distributed, containerized world • Kubernetes configs included • http://vitess.io/ Volume Volume Volume Volume Volume Vitess -
Using the same image for *almost* the exact same thing?
Often need changes based on context: ◦ dev, test, prod ◦ GCP, AWS, your servers ◦ feature flags ◦ logging config ◦ etc, etc... Environment variables! Configuration -
secret.yaml apiVersion: "v1" kind: "Secret" id: "mysql-pw" metadata: name: "mysql-pw"
data: password: "bXlzZWNyZXRwYXNzd29yZA==" Secrets pod.yaml <snip> volumes: - name: "password_volume" secret: secretName: "mysql-pw" containers: - name: "mysql" image: "gcr.io/google-samples/mysql:secret" volumeMounts: - name: "password_volume" mountPath: "/etc/mysql-password" readOnly: true bash (inside the container) $ cat /etc/mysql-password/password -
Watch John Wilkes talk • http://www.infoq.com/presentations/cluster-management-google Paper: Large-scale cluster management
at Google with Borg • http://research.google.com/pubs/pub43438.html Kubernetes: http://kubernetes.io Try Kubernetes quickly and easily on Google Container Engine • https://cloud.google.com/container-engine/ Try these patterns in Kubernetes • https://github.com/GoogleCloudPlatform/kubernetes/blob/master/docs/services. md#services-without-selectors • https://github.com/GoogleCloudPlatform/kubernetes/tree/master/examples/mysql- wordpress-pd • https://github.com/GoogleCloudPlatform/kubernetes/tree/master/examples/cassandra • https://github.com/GoogleCloudPlatform/kubernetes/blob/master/docs/secrets.md • https://vitess.io/ Next steps (take a photo)