Pinterest's Android CI Breakthrough: Runtime-Aware Sharding Slashes Build Times

Deconstructing Pinterest's CI Success

The key insight is the use of runtime-aware sharding, leveraging historical data to optimize test execution time distribution. This is noteworthy because it moves beyond simplistic sharding strategies, offering a more efficient and predictable CI pipeline. A limitation could be the initial setup cost and the need for a robust data collection and management system (Metro in Pinterest's case). However, the benefits in terms of reduced build times and increased developer productivity likely outweigh the implementation challenges. This approach benefits Android developers, DevOps engineers, and any team facing slow CI builds, particularly those with large and complex test suites.

Key Points

• Pinterest reduced Android CI build times by 36% using runtime-aware test sharding.
• They built an in-house platform (PinTestLab) with Android emulators on EC2 instances for better control.
• The core innovation is a sharding algorithm that groups tests by expected runtime, minimizing variance across shards.

---

📖 Source: Pinterest Engineering Reduces Android CI Build Times by 36% with Runtime-Aware Sharding