lsm-tree

A K.I.S.S. implementation of log-structured merge trees (LSM-trees/LSMTs) in Rust.

This crate only provides a primitive LSM-tree, not a full storage engine. For example, it does not ship with a write-ahead log. You probably want to use https://github.com/fjall-rs/fjall instead.

About

This is the most feature-rich LSM-tree implementation in Rust! It features:

• Thread-safe BTreeMap-like API
• 100% safe & stable Rust
• Block-based tables with compression support
• Range & prefix searching with forward and reverse iteration
• Size-tiered, (concurrent) Leveled and FIFO compaction
• Multi-threaded flushing (immutable/sealed memtables)
• Partitioned block index to reduce memory footprint and keep startup time short [1]
• Block caching to keep hot data in memory
• Bloom filters to increase point lookup performance (bloom feature, disabled by default)
• Snapshots (MVCC)
• Key-value separation (optional) [2]
• Single deletion tombstones ("weak" deletion)

Keys are limited to 65536 bytes, values are limited to 2^32 bytes. As is normal with any kind of storage engine, larger keys and values have a bigger performance impact.

Feature flags

lz4

Allows using LZ4 compression, powered by lz4_flex.

Disabled by default.

miniz

Allows using DEFLATE/zlib compression, powered by miniz_oxide.

Disabled by default.

bloom

Uses bloom filters to reduce superfluous disk I/O during point reads, improving performance, but also increasing memory usage.

Disabled by default.

Stable disk format

The disk format is stable as of 1.0.0.

2.0.0 uses a new disk format and needs a manual format migration.

Future breaking changes will result in a major version bump and a migration path.

License

All source code is licensed under MIT OR Apache-2.0.

All contributions are to be licensed as MIT OR Apache-2.0.

Development

Run benchmarks

cargo bench --features bloom --features lz4 --features miniz

Footnotes

[1] https://rocksdb.org/blog/2017/05/12/partitioned-index-filter.html

[2] https://github.com/facebook/rocksdb/wiki/BlobDB