sled 0.29.2

a modern embedded database
| key | value |
| :-: | --- |
| [documentation]( | [![documentation](]( |
| [chat about databases with us]( | [![chat](]( |
| [help us build what you want to use]( | [![Open Collective backers](]( |

<p align="center">
  <img src="" width="20%" height="auto" />

# sled - ~~it's all downhill from here!!!~~

A (beta) modern embedded database. Doesn't your data deserve a (beta) beautiful new home?

use sled::Db;

let tree = Db::open(path)?;

// insert and get, similar to std's BTreeMap
tree.insert(k, v1);
assert_eq!(tree.get(&k), Ok(Some(v1)));

// range queries
let mut iter = tree.range(k..);
assert_eq!(, Some(Ok((k, v2))));
assert_eq!(, None);

// deletion

// compare and swap
tree.compare_and_swap(k, Some(v1), Some(v2));

// block until all operations are stable on disk
// (flush_async also available to get a Future)

# performance

* 2 million sustained writes per second with 8 threads, 1000 8 byte keys, 10 byte values, intel 9900k, nvme
* 8.5 million sustained reads per second with 16 threads, 1000 8 byte keys, 10 byte values, intel 9900k, nvme

what's the trade-off? sled uses too much disk space sometimes. this will improve significantly before 1.0.

# features

* [API]( similar to a threadsafe `BTreeMap<[u8], [u8]>`
* fully serializable multi-key and multi-Tree [transactions]( involving up to 69 separate Trees!
* fully atomic single-key operations, supports [compare and swap](
* zero-copy reads
* [write batch support](
* [subscription/watch semantics on key prefixes](
* [multiple keyspace/Tree support](
* [merge operators](
* forward and reverse iterators
* a crash-safe monotonic [ID generator]( capable of generating 75-125 million unique ID's per second
* [zstd]( compression (use the `compression` build feature)
* cpu-scalable lock-free implementation
* SSD-optimized log-structured storage
* prefix encoded keys reducing the storage cost of complex keys

# a note on lexicographic ordering and endianness

If you want to store numerical keys in a way that will play nicely with sled's iterators and ordered operations, please remember to store your numerical items in big-endian form. Little endian (the default of many things) will often appear to be doing the right thing until you start working with more than 256 items (more than 1 byte), causing lexicographic ordering of the serialized bytes to diverge from the lexicographic ordering of their deserialized numerical form.

* Rust integral types have built-in `to_be_bytes` and `from_be_bytes` [methods](
* bincode [can be configured]( to store integral types in big-endian form.

# interaction with async

If your dataset resides entirely in cache (achievable at startup by setting the cache
to a large enough value and performing a full iteration) then all reads and writes are
non-blocking and async-friendly, without needing to use Futures or an async runtime.

To asynchronously suspend your async task on the durability of writes, we support the
[`flush_async` method](,
which returns a Future that your async tasks can await the completion of if they require
high durability guarantees and you are willing to pay the latency costs of fsync.
Note that sled automatically tries to sync all data to disk several times per second
in the background without blocking user threads.

# architecture

lock-free tree on a lock-free pagecache on a lock-free log. the pagecache scatters
partial page fragments across the log, rather than rewriting entire pages at a time
as B+ trees for spinning disks historically have. on page reads, we concurrently
scatter-gather reads across the log to materialize the page from its fragments.
check out the [architectural outlook](
for a more detailed overview of where we're at and where we see things going!

# goals

1. don't make the user think. the interface should be obvious.
1. don't surprise users with performance traps.
1. don't wake up operators. bring reliability techniques from academia into real-world practice.
1. don't use so much electricity. our data structures should play to modern hardware's strengths.

# known issues, warnings

* if reliability is your primary constraint, use SQLite. sled is beta.
* if storage price performance is your primary constraint, use RocksDB. sled uses too much space sometimes.
* quite young, should be considered unstable for the time being.
* the on-disk format is going to change in ways that require manual migrations before the `1.0.0` release!
* until `1.0.0`, sled targets the *current* stable version of rust. after `1.0.0`, we will aim to trail current by at least one version. If this is an issue for your business, please consider helping us reach `1.0.0` sooner by financially supporting our efforts to get there.

# plans

* Typed Trees that support working directly with serde-friendly types instead of raw bytes,
  and also allow the deserialized form to be stored in the shared cache for speedy access.
* [LSM tree]( write performance
  with [traditional B+ tree]( read performance
* MVCC and snapshots
* forward-compatible binary format
* concurrent snapshot delta generation and recovery
* consensus protocol for [PC/EC]( systems
* pluggable conflict detection and resolution strategies for gossip + CRDT-based [PA/EL]( systems
* first-class programmatic access to replication stream

# fund feature development and get commercial support

Want to support the project, prioritize a specific feature, or get commercial help with using sled in your project? [Ferrous Systems]( provides commercial support for sled, and can work with you to solve a wide variety of storage problems across the latency-throughput, consistency, and price performance spectra. [Get in touch!](

<p align="center">
  <a href="">
    <img src="" width="60%" height="auto" />

Want to support development but don't need commercial support? Help us out via [Open Collective](!

# special thanks

<p align="center">
  <a href="">
    <img src="" width="20%" height="auto" />

Special thanks to [Meili]( for providing engineering effort and other support to the sled project. They are building [an event store]( backed by sled, and they offer [a full-text search system]( which has been a valuable case study helping to focus the sled roadmap for the future.

<p align="center">
  <a href="">
    <img src="" width="20%" height="auto" />

Additional thanks to [Arm](, [Works on Arm]( and [Packet](, who have generously donated a 96 core monster machine to assist with intensive concurrency testing of sled. Each second that sled does not crash while running your critical stateful workloads, you are encouraged to thank these wonderful organizations. Each time sled does crash and lose your data, blame Intel.

Finally, thanks to everyone who helps out by contributing on [Open Collective](!

# contribution welcome!

want to help advance the state of the art in open source embedded
databases? check out [](!

# References

* [The Bw-Tree: A B-tree for New Hardware Platforms](
* [LLAMA: A Cache/Storage Subsystem for Modern Hardware](
* [Cicada: Dependably Fast Multi-Core In-Memory Transactions](
* [The Design and Implementation of a Log-Structured File System](