Updated on 2025-05-22 GMT+08:00

OPS04-02 Using Continuous Deployment Models

Continuous deployment models facilitate rapid fault detection, significantly reducing the impact on end users.

A popular model within this approach is canary deployment, where new functions are initially introduced to a limited set of internal or external users. Once the release is confirmed to be stable, it is gradually rolled out to larger groups until the entire user base has transitioned to the new version.

Another common model is blue-green deployment, which leverages two identical production environments. The blue environment serves all live traffic, while the green environment is updated with new features and tested internally. Once confirmed to be stable, production traffic is gradually shifted from blue to green, mirroring the controlled approach of canary deployment. When the transfer is complete, the green environment takes over as the primary production instance, and the former blue environment is prepped for future updates. The two instances are clearly separated to prevent faults across environments.

  • Risk level

    High

  • Key strategies

    When choosing a deployment model, it is essential to allocate sufficient time between each stage to monitor key workload health metrics. Avoid compressing deployments into minutes. Instead, extend the intervals to hours or even days. This pacing ensures users across various regions and tasks have ample time to interact with each version. Additionally, longer intervals can better accommodate differences in time zones and usage patterns.

  • Related cloud services and tools

    CodeArts Deploy