Docs: Scaling/HA page needs cleaning up
Problem to solve
Deploying GitLab can be complicated because there are so many options you can choose from environments (bare metal, cloud, kubernetes, etc) to scaling strategies. This puts a heavy burden on solution architects and TAMs for small customers who would be better served by some well-defined prescriptive installation architectures. On the scaling and high availability page we give a lot of options but no well-defined suggestions for various sizes of installation. @sytses and I recorded a video talking about some of the shortcomings of the page and suggesting some easy quick fixes (and some harder ideas).
Some details:
- There's a list of GitLab components that should also link to the pages about those components #205635 (closed)
- We inadvertently push customers to high availability features when they may not need them #205632 (closed)
- We talk a lot about all the various options, with no recommendations for specific strategies based on user count/repository sizes #205632 (closed) (same issue as above)
- In our reference architecture descriptions we give GCP node sizes; we should include equivalent AWS node specifics #205633 (closed)
- We need to be clear that gitaly only replaces NFS for repository data. There is still a requirement for shared storage, whether NFS or object storage. Probably move this section from the gitaly page to this page. #205634 (closed)
Edited by Mike Lewis