pebble VS moss

To improve the compacting speed of L0 files, we took inspiration from the design of the CockroachDB storage engine pebble.

https://github.com/cockroachdb/pebble ?

Pebble is a LevelDB/RocksDB-inspired key-value database focused on performance. PebbleDB is replacing Fantom’s previous RPC software to match Fantom’s sheer throughput. PebbleDB is consistently faster than previous options and it has helped the Fantom network to consistently reach synchronization with the network quicker while reducing the duration of API requests.

https://github.com/cockroachdb/pebble Pure go SSD native key-value store. You could think of it as map[[]byte][]byte on persistent storage.

Genji is a document-oriented, embedded, SQL database. It is build over Pebble which is a port of RocksDB in Go, by the authors of CockroachDB.

Out of curiosity, why write https://github.com/cockroachdb/pebble and not just use https://github.com/dgraph-io/badger?

There's the semi-recent pebble from the folks that make cockroach db. Haven't used it, but seems promising: https://github.com/cockroachdb/pebble

moss - Moss is a simple LSM key-value storage engine written in 100% Go.

I decided to dig through a database source code to answer that question. There are plenty of databases that use mmap. Some of them decided to not use anymore. Some examples: SQLite has an option of accessing disk content directly using memory-mapped I/O[1], it seems LevelDB used to use but it changed it[2], Lucene has an option with MMapDirectory[3], LMDB uses mmap[4], a simple key/value in-memory database from Counchbase called moss uses mmap for durability of in-memory data[5] and MongoDB removed mmap storage engine for WiredTiger[6].