review-performance

Performance regression detection in Apache Pinot query paths involves reviewing code diffs to surface per-row allocations, boxing, and synchronized blocks. It targets hot-path components like TransformOperator and FilterOperator to flag slowdowns before they reach production.

Common performance issues in Pinot segment scan and ForwardIndexReader diffs include per-row memory allocations, primitive boxing, synchronized blocks, and missing fast-paths. These inefficiencies in core per-row operators significantly degrade query throughput and scan latency.

Performance review findings for Pinot hot-path code changes are categorized using severity tags: CRITICAL, MAJOR, and MINOR. This severity-guided approach pairs concrete recommendations and verification steps with each flagged allocation or synchronization regression.

This code review process applies specifically to Apache Pinot per-row operator paths and related core areas. It targets TransformOperator, FilterOperator, ForwardIndexReader, and segment scanners to identify hot-path regressions rather than covering all Pinot components generally.

This hot-path review approach is not suited for Pinot PRs that do not modify query execution operators. It specifically analyzes diffs in per-row paths like TransformOperator, FilterOperator, and ForwardIndexReader to catch performance regressions in critical segment scan logic.