OCI — Load Testing and Scale Validation (issue 12 of 29)

Labels: ready-to-code
Depends on: Issue 1 (#4039 shared infra), Issue 9 (#TBD core connector)
RHIDP Stories: RHIDP-15268
Feature: RHDHPLAN-1507 — Epics RHIDP-15258, RHIDP-15294

Create load test harness generating 5,000+ AI asset entities, measure baseline vs with-AI-Catalog p95 latency (SLA: ≤10% degradation), validate processing-loop duration (entity emission → catalog API availability), and validate 2,000-image OCI sync completes within 5 minutes with peak memory under 500 MB.

Tasks

From openspec/changes/ai-catalog-entity-model/tasks.md group 7 (RHIDP-15268):

• 7.1 Create load test harness generating 5,000+ AI asset entities
• 7.2 Measure baseline p95 latency without AI Catalog entities
• 7.3 Measure with-AI-Catalog p95 latency with 5,000+ entities
• 7.4 Validate p95 latency degradation ≤10% SLA
• 7.5 Measure processing-loop duration
• 7.6 Document load test execution steps in tests/load/README.md

From openspec/changes/oci-skill-connector/tasks.md group 5 (RHIDP-15328):

• 5.1 Create mock OCI registry test harness with 2,000 synthetic skill images
• 5.2 Implement mock registry endpoints
• 5.3 Implement parallel manifest fetch with configurable concurrency
• 5.4 Implement batch processing (100 per batch)
• 5.5 Implement memory-bounded processing
• 5.6 Validate full sync of 2,000 images within 5 minutes
• 5.7 Validate incremental sync with 10% churn within 2 minutes
• 5.8 Log throughput and memory usage
• 5.9 Fail if throughput drops below 6 images/second
• 5.10 Verify peak memory under 500 MB

Specifications

• openspec/changes/ai-catalog-entity-model/specs/performance-resilience/spec.md
• openspec/changes/oci-skill-connector/specs/incremental-sync-scale/spec.md

Start with the two specifications in openspec/changes/ai-catalog-entity-model/specs/performance-resilience/spec.md and openspec/changes/oci-skill-connector/specs/incremental-sync-scale/spec.md, checking the dependencies on issues #4039 and #TBD. Implement the harnesses and document execution in tests/load/README.md. Done means meeting the stated latency, processing-duration, throughput, sync-time, churn, and 500 MB memory targets for the specified entity and image counts.