Extend Tenant Scale error budget exception
Description
This is a follow-up gitlab-com/www-gitlab-com!121190 (merged) to extend the Tenant Scale error budget exception for 6 months. Please read the previous merge request with all the context.
We are asking for an extension because we need time to start working on the error budgets while progressing on our schedule:
- The team is focused on Organization and Cells, with 2 borrow requests, and we are also hiring more engineers.
- Schedule identified error budget improvements gitlab-org&7832.
- We also want to explore more ideas:
- Review existing rate limits
- Ask authors of popular libraries/integrations that use these endpoints to use them more efficiently: gitlab-org/gitlab#387850 (closed), https://gitlab.com/gitlab-org/tenant-scale-group/group-tasks/-/issues/151, and https://gitlab.com/gitlab-org/tenant-scale-group/group-tasks/-/issues/78#note_1508766533.
- Consider using 17.0 to introduce breaking changes that would make the default API responses more efficient, such as defaulting to the
simple=true
payload or only showing projects that a user is a membermembership=true
of as opposed to all visible projects. - See if we can use ElasticSearch after all projects are indexed gitlab-org/gitlab#428070 (closed).
- Profile endpoints with the performance bar to get a flamegraph.
- Verify if budget measurement is impacted but not overall availability.
In order to do that, we need time to review, organize, and prioritize the work. That's the reason we are asking for a 6 months extension.