While supporting a vendor CI/CD product (Atlassian Bamboo) for ~4,000 developers, we noticed really bad performance causing Bamboo to lock users out during the day, cause teams to restart builds, over whelm queues, all bad news.

We did some heap dumps and ran profiling to identify a few significant queries that were taking a long time to run and getting hit rather often. This led us to a configuration page exposed to pipeline admins which ran an intensive backend query on every page load.

I was able to contrast this with the project summary page that loaded significantly faster. Why? – It used an index.

I submitted a PR (code change) to the Atlassian team which updated the admin page to use the more performant search index, and only invoke search when an admin attempts to change that field.

This change reduces page load from from 2+ minutes to nearly instant, with an incredibly responsive UI.. Because it used a cached index instead of live DB queries it also cut a significant amount of resources to be used by more important tasks like actually running builds!

https://bitbucket.org/atlassianlabs/bamboo-after-deployment-trigger/pull-requests/2/fixes-issue-2-eliminate/diff