Crates.io | fjall |
lib.rs | fjall |
version | 2.3.2 |
source | src |
created_at | 2023-12-21 17:17:56.379608 |
updated_at | 2024-11-03 16:26:01.075504 |
description | LSM-based key-value storage engine |
homepage | https://github.com/fjall-rs/fjall |
repository | https://github.com/fjall-rs/fjall |
max_upload_size | |
id | 1077261 |
size | 349,581 |
Fjall is an LSM-based embeddable key-value storage engine written in Rust. It features:
It is not:
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.
Like any typical key-value store, keys are stored in lexicographic order. If you are storing integer keys (e.g. timeseries data), you should use the big endian form to adhere to locality.
cargo add fjall
use fjall::{Config, PersistMode, Keyspace, PartitionCreateOptions};
// A keyspace is a database, which may contain multiple collections ("partitions")
// You should probably only use a single keyspace for your application
//
let keyspace = Config::new(folder).open()?; // or open_transactional for transactional semantics
// Each partition is its own physical LSM-tree
let items = keyspace.open_partition("my_items", PartitionCreateOptions::default())?;
// Write some data
items.insert("a", "hello")?;
// And retrieve it
let bytes = items.get("a")?;
// Or remove it again
items.remove("a")?;
// Search by prefix
for kv in items.prefix("prefix") {
// ...
}
// Search by range
for kv in items.range("a"..="z") {
// ...
}
// Iterators implement DoubleEndedIterator, so you can search backwards, too!
for kv in items.prefix("prefix").rev() {
// ...
}
// Sync the journal to disk to make sure data is definitely durable
// When the keyspace is dropped, it will try to persist with `PersistMode::SyncAll` as well
keyspace.persist(PersistMode::SyncAll)?;
To support different kinds of workloads, Fjall is agnostic about the type of durability
your application needs.
After writing data (insert
, remove
or committing a write batch/transaction), you can choose to call Keyspace::persist
which takes a PersistMode
parameter.
By default, any operation will flush to OS buffers, but not to disk.
This is in line with RocksDB's default durability.
Also, when dropped, the keyspace will try to persist the journal to disk synchronously.
!!! A single keyspace may not be loaded in parallel from separate processes.
However, Fjall is internally synchronized for multi-threaded access, so you can clone around the Keyspace
and Partition
s as needed, without needing to lock yourself.
For an async example, see the tokio
example.
Uses bloom filters to reduce disk I/O when serving point reads, but increases memory usage.
Enabled by default.
Allows using LZ4
compression, powered by lz4_flex
.
Enabled by default.
Allows using DEFLATE/zlib
compression, powered by miniz_oxide
.
Disabled by default.
Allows opening a transactional Keyspace for single-writer (serialized) transactions, allowing RYOW (read-your-own-write), fetch-and-update and other atomic operations.
Enabled by default.
Allows opening a transactional Keyspace for multi-writer, serializable transactions, allowing RYOW (read-your-own-write), fetch-and-update and other atomic operations. Conflict checking is done using optimistic concurrency control.
Disabled by default.
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.
For the underlying LSM-tree implementation, see: https://crates.io/crates/lsm-tree.
See here for practical examples.
And checkout Smoltable
, a standalone Bigtable-inspired mini wide-column database using fjall
as its storage engine.
How can you help?
All source code is licensed under MIT OR Apache-2.0.
All contributions are to be licensed as MIT OR Apache-2.0.