Soru

Zorluk: OrtaRelease Management and Deployment Strategies

An operations team is establishing an automated canary deployment process for a critical stateless application running on Google Kubernetes Engine (GKE) behind an External HTTP(S) Load Balancer. The deployment pipeline must minimize user risk by using progressive traffic routing and metrics validation. Arrange the steps in the correct chronological order to complete a safe canary release.

  1. 1Deploy the candidate release version alongside the existing stable release workload in the cluster.
  2. 2Configure traffic splitting on the backend service to route a small percentage (e.g., 5%) of live requests to the candidate deployment.
  3. 3Monitor Cloud Monitoring metrics against defined Service Level Indicators (SLIs) over a specified observation window.
  4. 4Progressively increase the traffic percentage allocated to the candidate release until it serves 100% of production traffic.
  5. 5Decommission the old stable workload resources and set the candidate version as the new baseline.

Cevap

The correct sequence for a progressive canary release is: 1) Deploy the candidate release version alongside the existing stable workload, 2) Configure traffic splitting on the backend service to route 5% of traffic to the candidate version, 3) Monitor Cloud Monitoring metrics against SLIs over an observation window, 4) Progressively increase traffic to 100%, and 5) Decommission the old stable workload resources.
A progressive canary release follows a strict sequence: first, the candidate version must be deployed in parallel without receiving traffic. Second, traffic splitting is introduced at a small ratio to limit the blast radius. Third, Cloud Monitoring metrics are evaluated over an observation window to ensure SLIs match SLO requirements. Fourth, traffic is incrementally shifted to 100% upon successful evaluation. Finally, legacy workloads are decommissioned.

Adım Adım Çözüm

1
Provision candidate workload pods
Candidate release running and healthy in parallel with stable workload
Workload pods must be active and passing health checks prior to accepting live requests.
2
Initialize traffic split
5% of production traffic routed to candidate pods, 95% remains on stable pods
Limits blast radius in case the candidate release contains undetected regressions.
3
Evaluate operational telemetry
SLIs confirmed within SLO error budget boundaries
Automated verification ensures candidate stability before committing more user traffic.
4
Complete traffic migration
100% of live traffic routed to candidate release
Gradually scaling traffic completes the transition once stability is validated.
5
Clean up baseline resources
Old stable deployment deleted; candidate becomes new baseline
Frees compute capacity and avoids configuration drift.

Anahtar Kavram

Progressive Canary Release and Traffic Splitting Management
Bu soruyu puanla