Deployment Strategies

Deployment Strategies

• It is a way to change or upgrade an application

Rollout:

Update server with new version

Rollback:

Revert recently updated servers back to previous

Recreate:

• Terminate all running instances and recreate with new version.

• Users experience downtime

• Can be fast * simple

• Rollback is not possible

• Ideal for non-production workloads

spec:
    replicas: 3
    strategy:
        type: Recreate

Rolling Updates:

• Rolling updates slowly replace pod one by one

• This is default strategy of Kubernetes

• Reduced availability might happen

• Rollback are slow and hard

• Deploys will be slow

spec:
    replicas: 3
    strategy:
        type: RollingUpdate
        rollingUpdate:
            maxSurge: 2
            maxUnavailable: 0

Rolling update

Canary:

• Deploy a new version of app into new pod and serve it to a subset of existing users

• If no-bug then replace old pods

• Fast rollout

• Slow rollback

• No drop in availability

Blue Green:

• Completely create a new environment of all compnents

• Send traffic to new environment if okay then terminate old (blue)

• Gree are new env

• Blue are old env

• Zero downtime

• No reduced availability

• Slow to deploy but faster than canary

• Instantly rollback to previous infra

Blue Green

A:B testing or Red/Black

• Uses canary or Blue Green method of deployment but serves the new app to subset of users based on a set of load balancing rules

• Intentiion for long-period of time

A:B testing

Dark Launches:

• Feature on a subset of users

• Turn neew feature on and off

Dark Launches

Commands:

kubectl rollout history deploy/<name>

kubectl rollout status deploy/<name>

kubectl rollout undo deploy/<name>