Apache, nginx, Docker and the Swear Words We Love
summary
An Italian talk reports performance troubleshooting during upgrades from Laravel 8 to 11 and PHP 7 to 8 when moving to Docker in production: enabling OPcache restored and improved throughput (from ~280 to ~670–800 req/min), while a non-Docker AMI reached ~850 req/min. Redis failures caused by an Elasticache DNS issue were fixed by using persistent Redis connections (connections fell from ~4000 to ~20). Web server and runtime comparisons include Apache vs nginx (1594 vs 1642 req/min), Octane+FrankenPHP with 15 workers (~1719 req/min), and a FrankenPHP static build that reduced API latency from 167 ms to 27 ms.
AI-written summary · about
chapters (13)
- 0:00Introduction
- 3:01Why we started with deployments
- 5:52Laravel 8 to 11 migration disaster
- 8:59Docker performance issues
- 11:05OPcache disabled
- 13:16Redis bottleneck and DNS problem
- 15:25Troubleshooting for stable performance
- 15:46Apache vs Nginx performance
- 19:37Git JIT tests poor results
- 22:04Octane and FrankenPHP solution
- 26:19FrankenPHP static build fast response
- 29:36Final ranking and Docker thoughts
- 31:46Q&A
related talks
58mMassively Scaled High Performance Web Services with PHP
37mScaling LAMP doesn't have to suck
47mFrom Legacy to High-Load: Evolution of a Web Application
31mFrom Runtime to Resilience: Scaling PHP
32mWordPress Scalability
39mPHP Performance Demystified - Measure, Understand, Optimise
43mnginx for fun and performance
45mNginx: The power within