Operations

This section is for running RecSys in production: performance, readiness, and on-call runbooks.

Who this is for

• SREs and on-call engineers running RecSys in production
• Engineering teams sizing capacity and validating production readiness

What you will get

• A production readiness checklist and baseline benchmarks
• Failure-mode diagnosis and safe remediations (with runbook links)
• The first pages to open when the service is not ready or recommendations go empty

Quick paths

• Performance & capacity
  Sizing guidance and performance expectations.
• Baseline benchmarks
  Reproducible “anchor numbers” and a template to record your own runs.
• Production readiness checklist
  Pre-flight checks before you go live.
• Failure modes & diagnostics
  Common symptoms, likely causes, and safe fixes (with links to runbooks).
• Service not ready (runbook)
  Triage steps when the API fails readiness.
• Empty recs (runbook)
  Common causes and safe remediations.
• Pipelines runbooks
  Day-2 operations for the offline layer.

Read next

• Production readiness checklist: Production readiness checklist (RecSys suite)
• Failure modes & diagnostics (with runbooks): Failure modes & diagnostics (baseline)
• Pipelines SLOs and freshness: SLOs and freshness