One Box Doesn't fit all
summary
Conference recap: claims, numbers, tools and outcomes. Flysystem hits near 10 million installs and a regex bug causes 200 million operations that brought Stack Overflow down. A 40 MB file test on a tiny Vagrant box with 100 concurrent requests shows ~20% failures and ~3 req/s when reading into memory; switching to streams removes failures, triples throughput and shrinks latency variance by ~10x. Practical fixes shown: nginx load‑balancer config, Varnish request coalescing, cache warming, Redis/S3 for stateless apps, Ansible/Prometheus monitoring, and an Elasticsearch memory fix that stabilizes clusters.
AI-written summary · about
chapters (16)
- 0:00Introduction to scaling
- 0:57The problem with scale
- 4:42Resource management
- 7:07Understanding resource bounds
- 10:46Memory optimization example
- 14:21Improving stability
- 17:04Effective monitoring data
- 18:43The role of caching
- 24:09Pragmatic caching solutions
- 25:05Vertical scaling risks
- 26:59Horizontal scaling approach
- 30:21Automation and maintenance
- 32:07Monitoring and status pages
- 33:43Stateless application design
- 35:48Strategic optimization
- 39:10Load balancing details
related talks
31mBobby Bouwmann :: Surviving Large Applications
44mBuilding a complex product while scaling a team and a business on Laravel
34mFreek Van der Herten :: Uncharted Packages
28mChristoph Rumpel :: Managing 5000+ Tests Efficiently
29mEffective tests and testing effectively
30mTim Macdonald :: Thoughtful Performance
25mLess Pain with Gain, how Laravel helps you scale your Applications
28mDaniel Coulbourne :: Flip the Switch Slowly