Catalogue / uber-2024-cachefront

Uber integrates CacheFront on Redis into Docstore to serve over 40 million reads per second across all Docstore instances

Uber, 2024. CacheFront, integrated Redis caching layer for Docstore.

Organisation
Uber
Evidence
Reported from production
Source
How Uber Serves Over 40 Million Reads Per Second from Online Storage Using an Integrated Cache

Uber Engineering Blog, published 2024-02-15

Records

Decision#d1

Integrate opt-in Redis caching into Docstore's query engine instead of per-team caches or scaling disk storage

Decision#d2

Invalidate cached rows via MySQL binlog CDC and atomic Lua timestamp checks, plus an explicit point-write API

Decision#d3

Warm remote-region caches by replicating only Redis keys and triggering local query-engine reads on miss

Decision#d4

Shard multiple Redis clusters per Docstore instance using a scheme distinct from database sharding

Decision#d5

Protect read tail latency from degraded Redis nodes with sliding-window circuit breakers and P99.99 adaptive timeouts

Sign in to read what was chosen, what was rejected and what it cost, each with its place in the source. An account is free.