badger

Shape: LSM with separate value log, pure Go, transactions Repository: github.com/dgraph-io/badger

badger is an LSM that stores keys and values separately: keys go in the LSM tree, values in a companion log. That split makes writes cheap and keeps the tree small, at the cost of disk space and slower ordered scans.

It is the most well-rounded writer in this set: fastest fresh writes on the laptop, the lowest tail latency on a mixed read-update workload, and one of only two engines here that batch durable commits.

Best at

• Fresh writes on a laptop or a few cores. 239,000 writes/sec on the M4, top of the ingest table.
• Durable writes. 16,000 durable writes/sec through group commit, 20x the per-commit engines. See durable writes.
• Low tail on a mixed load. 215 us p99 on the 50/50 mixed workload, the tightest there.

Watch out for

• Disk footprint. 22x the raw data after a write-heavy run, the largest here, because the value log reclaims dead space lazily. It comes back as GC runs, but provision for the peak.
• Slow ordered scans. 29,000 keys/sec, far behind the other LSMs, because an ordered key scan keeps jumping to the value log.
• Write speed drops on servers. On the Linux hosts pebble overtakes it; badger's value-log GC does not scale across cores the way pebble's compaction does.

Reach for it when

You write a lot, you need those writes durable under concurrency, and you have disk to spare. For the same write profile with a small footprint on a server, compare pebble.