To optimize our frontend performance, I wanted to identify which API endpoints degraded, trace to slow dependencies, and identify the code changes responsible.
What makes this hard?
Your APM shows slow traces, but doesn't show you the code. GitHub shows you code, but doesn't show you performance. Git history shows you changes, but doesn't connect to live system behavior. So debugging requires manual work across 5 disconnected tools to:
- Query your APM to compare performance and find the slowdown
- Dig through trace spans to find the slow database query
- Search your codebase to locate where that query runs
- Check git history to find who changed it and when
- Manually connect: traces → SQL → code location → git commit → performance timeline
How did Resolve AI help?
With one query, Resolve AI automatically queried Tempo traces to compare week-over-week performance across endpoints:
- Identified the slow dependency: Ad service database queries showed 85% increase in latency (117ms → 216ms average), with worst cases reaching 17 seconds
- Found the specific SQL query: select count(?) from product executing on PostgreSQL database, affecting 89.2% of queries
- Located exact code location: AdService.java line 219 in getAds() method
- Retrieved git history: Commit 80dd1d5 from June 17, 2025
- Recommended the fix: Add connection pooling (HikariCP), cache the count value, make query async, or remove entirely since count doesn't appear used in ad selection logic
Resolve AI connected trace evidence (216ms average query time, 17s worst case) to exact code location (line 219, getAds() method) to git commit (80dd1d5, June 17) to root cause analysis (no connection pooling, synchronous blocking). The investigation revealed that a seemingly simple database query added months ago was creating a hidden performance bottleneck affecting nearly 90% of ad service requests.
