Managing Service Deployment

1. Using Blue-Green Deployment

StepDetail
Two envsBlue (live) + Green (new)
SwitchFlip LB to green
RollbackFlip back instantly
Cost2× capacity during cutover

2. Implementing Canary Releases

StepDetail
Route %1% → 10% → 50% → 100%
Health gatesAuto promote/abort on metrics
ToolsArgo Rollouts, Flagger

3. Using Rolling Updates

ParamDetail
maxSurgeExtra pods during update
maxUnavailablePods that may be down
Readiness probeRequired for safe rolling

4. Implementing Feature Toggles

TypeUse
ReleaseHide unfinished work
ExperimentA/B
OpsKill switch
PermissionPer-user gating
ToolsLaunchDarkly, Unleash, Flagsmith

5. Using A/B Testing

ElementDetail
HypothesisMetric to move
BucketingSticky per user
Statistical significanceSample size up front

6. Implementing Shadow Deployment

AspectDetail
Mirror trafficTo new version; ignore response
CompareLatency, error, output
RiskSide-effects (writes) — read-only safe

7. Managing Deployment Pipelines

StageDetail
BuildCompile, unit, SAST/SCA
TestIntegration, contract, security
StageSmoke, perf
ProdCanary → full
ToolsGitHub Actions, GitLab CI, Argo CD

8. Implementing Rollback Strategies

MethodDetail
Image revertRedeploy previous tag
GitOpsgit revert → reconcile
SchemaBackward-compatible migrations
Forward-fixSometimes safer

9. Using Infrastructure as Code

ToolScope
Terraform / OpenTofuCloud + multi-provider
PulumiReal code
CloudFormation / CDKAWS
CrossplaneCloud via K8s

10. Implementing Zero-Downtime Deployment

RequirementDetail
Backward-compat schemaExpand → migrate → contract
Graceful shutdownDrain conns, finish in-flight
Readiness probesGate traffic
Connection drainingLB awareness

11. Managing Database Migrations

StepDetail
ExpandAdd new col/table; nullable
MigrateBackfill in batches
ContractDrop old col after deploy
ToolsFlyway, Liquibase

12. Using Deployment Tools

ToolUse
Argo CD / FluxGitOps
Argo Rollouts / FlaggerProgressive delivery
Helm / KustomizeK8s manifests
SpinnakerMulti-cloud pipelines