At Dgraph, we’re building the most advanced graph database in the world. It does distributed transactions, low-latency arbitrary depth joins, traversals, provides synchronous replication and horizontal scalability — with a simple GraphQL-like API. Does it sounds like a tough problem? That’s because it is. This blog post is about Badger, the key-value database that makes it all happen under the hood, housing all Dgraph data, including Raft logs. Badger uses an LSM tree structure for holding keys.

We have built an efficient and persistent log structured merge (LSM) tree based key-value store, purely in Go language. It is based upon WiscKey paper included in USENIX FAST 2016. This design is highly SSD-optimized and separates keys from values to minimize I/O amplification; leveraging both the sequential and the random performance of SSDs. We call it Badger. Based on benchmarks, Badger is at least 3.5x faster than RocksDB when doing random reads.