FFT-Performance - Profiling and Optimization
Overview
Section titled “Overview”FFT-Performance is data-driven. It measures first, changes second. It identifies real bottlenecks through profiling and production telemetry, not speculation, and it establishes performance budgets that keep regressions from sneaking back in after each optimization.
Capabilities
Section titled “Capabilities”- Profiling: CPU, memory, allocation, flame graphs across Node.js, Python, Go, and JVM runtimes.
- Caching strategy: multi-tier caches, cache invalidation patterns, cache-aside vs write-through.
- Database optimization: index tuning, query plan analysis, connection-pool sizing.
- Load testing: k6, Locust, Gatling; realistic scenarios grounded in production traffic shape.
- Frontend performance: Core Web Vitals, bundle analysis, critical-path rendering, image optimization.
- Memory management: leak detection, GC tuning, heap snapshots, object-retention analysis.
- Scaling strategies: horizontal scaling triggers, autoscaling policies, queue-based load leveling.
- Performance budgets: enforced in CI so regressions fail builds, not production.
When to Use
Section titled “When to Use”- A feature is slow in production and you need root cause, not guesswork.
- Planning capacity for an expected traffic spike or growth period.
- Hitting Core Web Vitals targets on a critical user flow.
- Establishing performance budgets and CI guardrails for a new service.
Example Prompts
Section titled “Example Prompts”"Profile the /search endpoint under 500 concurrent users and identify the top three bottlenecks""Design a multi-tier caching strategy for the product catalog with realistic invalidation rules""Set up k6 load tests and CI performance budgets for the checkout flow"