High-Load Systems: Challenges in Social Network Development
summary
Numbers: 100 million users per month, 100 million posts per day and 9 billion post views per day drive the decisions. The system migrates from MySQL to dozens of custom C/C++ embedded databases and eight giant clusters, adds dynamic image resizing and WebP-like formats, and compiles PHP to KPHP for 7–10× runtime gains. Targets include five‑n uptime (~5 minutes downtime/year), zero data loss, heavy precomputation and caching, and sharded master‑slave replication; these changes measurably reduce latency and backend load.
AI-written summary · about
chapters (9)
- 0:00intro
- 0:14high load systems
- 2:00defining high load systems
- 4:02challenges in managing high load systems
- 4:47risks and limitations of external solutions
- 7:18case study: vk social network
- 10:13optimizing application code and architecture
- 14:09industry examples and best practices
- 16:03conclusion and key takeaways
related talks
52mPHP at the Firehose Scale
45mScaling: from 0 to 20 million users
50mBuilding and Operating Online Shops under Extreme Load
47mFrom Legacy to High-Load: Evolution of a Web Application
47mScaling: from 0 to 25 million users
32mWordPress Scalability
38m180,000 requests per second explained simply
11mDatabase Performance in High Traffic Environments