Harald Ng

Building distributed RocksDB with OmniPaxos in 8 minutes

In this 8-minute blog post, we show how to transform a simple single-server RocksDB service into a distributed version with multiple servers replicating the same RocksDB state. We use OmniPaxos as a replicated changelog that all servers execute according to guarantee consistency. We also show how to prevent the log from growing infinitely by using the snapshot feature in OmniPaxos.

Posted on June 25, 2023 by Harald Ng  ‐  8 min read