Memory: Amplifying SaaS Reliability Focus for Q3FY22

This is the Memory-specific issue for #2881 (closed). The purpose is to provide a tracking and organization issue for surfacing the critical customer needs and opportunities to improve SaaS reliability for each group. Once collected, we can surface this back up to the main parent issue in a clear way.

Asks

See: #2881 (closed)

  • PM: Understand and share what are the minimal-must-do features/capabilities we should deliver to meet customer ARR commitments from 14.2 through 14.6
  • EM: Understand and share What are the top most things you could advocate for to improve reliability of your areas. In addition to improving reliability, consider observability needs to detect and fix reliability issues when they do occur in SaaS

Discussion

For Memory, this is likely to be an interesting discussion. The key here is reliability, not necessarily performance.

SaaS Reliability items:

In current priority order

  1. Shifting Sidekiq workers to read replicas: gitlab-org&5691 (closed)
  2. Infra Feature Flags
  3. Sidekiq Metrics
  4. LB Improvement
  5. Remove Gitaly-ruby

Work in-flight that could be paused if needed

Edited by silv