Enable/gather metrics - latency, memory, cpu, etc.

What

As discussed in profiling approach GET environments provide optional prometheus deployments for exporting observability metrics, for example the 2k testbed metrics dashboard. By comparative monitoring with our environment deployments we should have baseline metrics to compare against.

We need to extract baselines for target architectures and identify the headroom we have to work with.

For example (made up numbers), we use 800/1000mb of memory on a 2k architecture, giving us 200mb of headroom. Bearing that in mind we identify

1. addtl memory usage for traffic profiles matching k6 load tests
2. thresholds for exceeding usage (presumably we want to stay under, say, 90% utilization)

Tasks

• Pull comparative metrics for GitLab Performance Tool automated reference environments including the following:
  • ✅ Overall score | thread
  • ✅ Time-to-first-Byte (TTFB) for git push events: avg and p90 | thread
  • ✅ CPU usage: avg and p90 | thread
  • ✅ Memory usage: avg and p90 | thread
  • ❌ Network traffic transmit (Mb/s): avg and p90 | thread
• Define target resource constraints

NOTE: All metrics should match traffic profile as documented in Reference Architecture scenarios; i.e. 60s_40rps for 2k.