Blue‑Green, Rolling, and Canary Deployment Strategies Overview

During project iteration, releasing new versions inevitably requires deployment, which brings risk; various deployment techniques exist, and this article summarizes the most common ones.

Blue‑Green Deployment

Blue‑green deployment uses two parallel environments: the current production (green) and a new version (blue). The blue environment is used for pre‑release testing without affecting users. After validation, traffic is switched to blue, which then becomes the new green after the old green is retired.

Characteristics

Reduces downtime during release.

Allows quick rollback to the previous version.

Considerations

Works best for loosely coupled, cohesive systems; complex systems may require data synchronization.

Database handling and transaction continuity must be planned.

Requires infrastructure support; non‑isolated environments (VM, Docker) may risk destroying one of the environments.

Rolling Deployment

Rolling deployment updates a subset of servers at a time, repeatedly replacing instances until the entire cluster runs the new version, which saves resources compared to blue‑green.

Characteristics

More resource‑efficient; only a fraction of instances are updated per cycle.

Rollback can be difficult.

Precautions

During the switch, unfinished and new business must be handled carefully, especially database compatibility.

Mixed microservice and monolithic architectures may cause service interruptions if not coordinated.

Database migration/rollback must be planned.

Infrastructure must support seamless updates.

Non‑isolated environments risk accidental destruction of instances.

Version incompatibility may appear temporarily, requiring compatibility handling.

Canary (Gray) Deployment

Canary deployment (also called gray release) gradually shifts traffic from the old version to a new one, often using a small subset of servers with reduced weight, monitoring performance before full rollout.

Steps

Prepare build artifacts, test scripts, configuration, and deployment manifests.

Deploy the canary server and test.

Remove the canary from the load balancer.

Upgrade the canary application, isolating traffic.

Run automated tests.

Re‑add the canary to the load balancer after successful verification.

If successful, upgrade remaining servers; otherwise, roll back.

A/B Testing

A/B testing differs from deployment strategies; it evaluates the actual effect of multiple already‑stable versions (e.g., conversion rate, order volume) by exposing different user groups to each version and analyzing results.

Key point: A/B testing focuses on performance metrics rather than release stability.

Additional Promotional Content (Non‑Academic)

The remainder of the source contains promotional messages for a ChatGPT community, giveaways, and links to other resources, which are not part of the technical discussion.

最近面试BAT，整理一份面试资料《Java面试BAT通关手册》，覆盖了Java核心技术、JVM、Java并发、SSM、微服务、数据库、数据结构等等。获取方式：点“在看”，关注公众号并回复 手册 领取，更多内容陆续奉上。