system-design

Estimate average QPS as DAU multiplied by actions per user divided by 86,400 seconds, then multiply by 2-5x for peak. Storage equals records per day times record size times retention period, rounded to order of magnitude.

Use base62-encoded auto-increment IDs stored in a key-value store, with cache-aside Redis for hot URLs. Choose 302 redirects for analytics or 301 for maximum caching performance, and estimate roughly 116 write QPS and 5,800 read QPS at 100M DAU.

Choose SQL when you need ACID transactions, complex joins, and a stable schema. Choose NoSQL for flexible schemas, simple key lookups at massive scale, or very high write throughput such as time-series data in wide-column stores.

Shard only after exhausting vertical scaling, read replicas, caching, and query optimization. Sharding fits write-heavy workloads or datasets too large for one server, using hash-based, range-based, or directory-based strategies with a high-cardinality shard key.

Fanout-on-write pre-computes feeds at post time, giving fast reads but expensive writes for celebrity accounts. Fanout-on-read assembles feeds at read time, giving cheap writes but slow reads. A hybrid approach pushes for normal users and pulls for celebrities.

Estimation targets order of magnitude, not precision, so it cannot replace load testing or real traffic measurement. Common mistakes include ignoring peak factors, media storage, replication overhead, and future growth rates.