Overview
Performance code reviews fail when they check code that works correctly in development without asking "what does this look like with 1 million rows?" A query with a full table scan takes 2ms on a development database with 1,000 rows. The same query takes 45 seconds on a production database with 50 million rows — and holds a lock during that time. Performance problems that are invisible in development are the most dangerous class of bug because they pass all tests and only surface under production load.
The Performance-Focused Code Review Framework identifies the specific code patterns that degrade at scale — database access patterns, algorithmic complexity, memory accumulation — and quantifies the expected impact at production data volumes.
Key Features
Three N+1 pattern types (explicit, hidden, ORM lazy)
Unbounded query patterns with fixes
O(n²) patterns with Map-based O(n) alternatives