#key-value-database #key-value #database-server #embedded-database #server #client

bin+lib yedb

Rugged crash-free embedded and client/server key-value database

47 releases

0.4.15 Mar 3, 2024
0.4.14 Dec 29, 2023
0.4.13 May 31, 2023
0.4.12 Mar 25, 2023
0.0.3 Feb 24, 2021

#128 in Database interfaces

Used in 2 crates


4.5K SLoC

yedb - rugged crash-free embedded and client/server key-value database (Rust implementation)

Cargo crate



  • client-sync synchronous client
  • client-async asynchronous client
  • cli yedb-cli
  • server yedb-server


  • Is it fast?

  • Rust version is pretty fast, except writes are still slow if auto-flush is enabled.

  • Is it smart?

  • No

  • So what is YEDB for?

  • YEDB is ultra-reliable, thread-safe and very easy to use.

  • I don't like Rust

  • There are other implementations.

Power loss data survive demo


YEDB is absolutely reliable rugged key-value database, which can survive in any power loss, unless the OS file system die. Keys data is saved in the very reliable way and immediately flushed to disk (this can be disabled to speed up the engine but is not recommended - why then YEDB is used for).

Rust version features

  • Rust version is built on top of Serde framework.

  • All key values are serde_json::Value objects.

  • Storage serialization formats supported: JSON (default), YAML, MessagePack and CBOR.

  • As byte type is not supported by serde_json::Value at this moment, Rust version can not handle byte key values.

  • Contains: embedded library, async server and command-line client (TCP/Unix socket only).

  • The command-line client is very basic. If you need more features, use yedb Python CLI.

Additional features over YEDB specs

  • "delete" command does not delete keys, they are moved to .trash folder instead

  • .trash folder is being cleaned when "purge" method is invoked.

  • "auto_bak" property tells server to automatically create backup key versions when key data is modified

  • "bak" keys are hidden


Binaries available at the releases page.

Run server:

./yedb-server /tmp/db1

Use client:

# get server info
./yedb-cli info
# set key value
./yedb-cli set x 5 -p number
# list all keys
./yedb-cli ls /
# edit key with $EDITOR
./yedb-cli edit x
# get key as JSON
./yedb-cli get x
# get help for all commands
./yedb-cli -h

Code examples

The database/client objects can be safely shared between threads using any kind of Lock/Mutex preferred.

Embedded example

use yedb::Database;
use serde_json::Value;

let mut db = Database::new();
let key_name = "test/key1";
db.key_set(&key_name, Value::from(123_u8)).unwrap();
println!("{:?}", db.key_get(&key_name));

TCP/Unix socket client example

use yedb::YedbClient;
use serde_json::Value;

let mut client = YedbClient::new("tcp://");
let key_name = "test/key1";
client.key_set(&key_name, Value::from(123_u8)).unwrap();
println!("{:?}", client.key_get(&key_name));

Async TCP/Unix socket client example

use serde_json::Value;
use yedb::{YedbClientAsync, YedbClientAsyncExt};

async fn test() {
    let mut client = YedbClientAsync::new("tcp://");
    let key_name = "test/key1";
    client.key_set(&key_name, Value::from(123_u8)).await.unwrap();
    println!("{:?}", client.key_get(&key_name).await);

Async TCP/Unix socket client pool example

use serde_json::Value;
use std::sync::Arc;
use std::time::Duration;
use yedb::{YedbClientPoolAsync, YedbClientAsyncExt};

async fn test() {
    let pool = Arc::new(
    let mut futs = Vec::new();
    for i in 0..10 {
        let task_pool = pool.clone();
        let fut = tokio::spawn(async move {
            let mut client = task_pool.get().await;
            let key = format!("test/key{}", i);
            client.key_set(&key, Value::from(i)).await.unwrap();
            println!("{}", client.key_get(&key).await.unwrap());
    for fut in futs {



Some benchmark data

  • CPU: Intel Core i7-8550U (4 cores)
  • Drive: Samsung MZVLB512HAJQ-000L7 (NVMe)
  • auto_flush: false
  • connection: Unix socket
  • server workers: 2
  • client threads: 4
set/number: 8164 ops/sec
set/string: 7313 ops/sec
set/array: 7152 ops/sec
set/object: 5272 ops/sec

get/number: 49709 ops/sec
get/string: 33338 ops/sec
get/array: 31426 ops/sec
get/object: 11654 ops/sec

get(cached)/number: 122697 ops/sec
get(cached)/string: 61206 ops/sec
get(cached)/array: 59309 ops/sec
get(cached)/object: 34583 ops/sec

increment: 7079 ops/sec


~564K SLoC