10 parts · 10 chapters
Deployment Techniques
Deploying is where most changes meet users and where most incidents start. The techniques in this course all do one thing: limit how many users a bad change can reach, and how fast you can take it back.
Ten parts: from scp to pipelines; rolling, recreate and blue-green; canaries with automated analysis; shadow traffic, dark launches and flag releases; database migrations during deploys; GitOps and promotion; deploying without Kubernetes; serverless and edge deploys; mobile and desktop releases; and rollback, roll-forward and release engineering as a role.
strategiesRecreate, rolling, blue-green, canary, shadow, dark launch, flags.
analysisAutomated canary analysis with Argo Rollouts and Flagger.
dataExpand and contract, online DDL, backfills during deploys.
GitOpsArgo CD and Flux, promotion between environments.
other platformsKamal, Fly, Render, ECS, Nomad, serverless, edge, app stores.
recoveryRollback versus roll-forward, release trains, release engineering.
00
From scp to Pipelines
The history and the vocabulary
1 ch · ~8 min01Rolling, Recreate and Blue-Green
Three replacement strategies
1 ch · ~8 min02Canary Releases and Automated Analysis
Canaries that judge themselves
1 ch · ~8 min03Shadow Traffic, Dark Launches and Flag Releases
Running new code where nobody can see it
1 ch · ~8 min04Database Migrations During Deploys
Compatible across two versions
1 ch · ~8 min05GitOps and Promotion Between Environments
Pull-based delivery
1 ch · ~8 min06Deploying Without Kubernetes
Platforms and their strategies
1 ch · ~8 min07Serverless and Edge Deploys
Versions, aliases and global propagation
1 ch · ~8 min08Mobile and Desktop Releases
Shipping to devices you do not control
1 ch · ~8 min09Rollback, Roll-Forward and Release Engineering
Rollback or roll forward
1 ch · ~8 minBuilt on Infra, SRE and the Production StackInfrastructure part 4 introduced CI/CD and progressive delivery; SRE part 5 covered release safety and budget gates. This course goes technique by technique.