Build GET Environment(s) For Performance Profiling

Tasks

At a high level, we will deploy ref-architecture spec'ed GET environments containing our patches to block on detection. This will allow us to run the corresponding test cases against the environments, measure TTFB and resource usage, and compare against baselines to determine whether profiles are within acceptable parameters.

• Determine size needed for GET (GitLab Environment Toolkit) environment - a 2k ref architecture will give us a Gitaly node separate from the Rails node, to simulate the latency we will see in most Production environments (Ross)
• Set up AWS GET environment with custom Omnibus build (Ross)
• Switch to GCP GET environment to make access/sharing easier for the team (Ross/Lucas)
• Share GET config(s) in central location (Ross) https://gitlab.com/gitlab-org/secure/pocs/gitlab-environment-toolkit-configs/-/merge_requests/4+
• Playbook to show us how to get reference architectures stood up efficiently Add runbook for Pre-receive secret detection (gitlab-com/content-sites/handbook!2745 - merged)
• Follow advanced monitoring setup. Prometheus is enabled on our GET instance, but we need to enable Grafana. This thread will be a good place to start: #428499 (comment 1629215984)
• Configure separate GCP GET instance for running golang PoC (Ross)