Using Skylight to Solve Real-World Performance Problems
summary
Skylight shows median and 95th percentile latencies rather than averages and pinpoints slow endpoints by combining slowness with popularity so fixes matter. It detects N+1 queries and surfaces the exact SQL so contributors change includes and submit PRs. Teams use explain analyze to prove a missing index helps Postgres and adopt dedicated index for the queried column. Investigation of memcached tiers leads to flushing and 100% cache hit rate after switching to collection caching.
AI-written summary Β· about
chapters (9)
- 0:00Start
- 1:33Referral Code
- 6:01Skylight Dashboard
- 6:49The Endpoints List
- 10:15What Is an N plus 1 Query
- 16:46Summary of What We Have Learned
- 19:44Code Triage
- 20:41Home Page
- 23:51Collection Caching
related talks
36mHow to Build a Smart Profiler for Rails
32mRails Performance First Aid
41mOptimizing your app by understanding your PostgreSQL database
37mMonitoring the Health of Your App
39mResolving Outages Faster with Better Debugging Strategies
27mLatency SLOs Done Right
28mResolving Outages Faster With Better Debugging Strategies
35m"10 performance sins of Rails applications"