Benchmark and performance test the Container Registry garbage collection algorithm on GCS
Problem to solve
In order to better understand how to optimize the performance of the GitLab Container Registry, we must first understand which parts of the algorithm take the longest to run and how each process works, so that we can better understand how to proceed wit the optimization effort.
Intended users
Further details
Proposal
Create and share performance benchmarks and optimization recommendations for the garbage collection algorithm on GCS, which will be used to help determine a path toward optimization.
Documentation
- Results can be shared in this issue or the epic, &2552 (closed)
Testing
What does success look like, and how can we measure that?
- Success looks like we have a good understanding of the bottlenecks in the algorithm and suggestions for how we may improve them.
Links / references
- Similar issue for S3: #37960 (closed)
Edited by Tim Rizzi