ClauDB

ClauDB is a REDIS implementation in Java. At the moment is in development and only implements a small subset of commands and features. The objetive is implement a full functional one-to-one replacement for REDIS (2.8 branch).

You will probably wonder why I do this, the answer is I do it Just For Fun.

Why ClauDB?

Initially I called this project TinyDB, but there's another project with the same name, so, I've decided to chage to ClaudDB.

Clau is 🔑 in Valencià, a language spoken in eastern Spain, and ClauDB is a key-value database.

Implemented commands

Server

FLUSHDB
INFO
TIME
SYNC
SLAVEOF
ROLE

Connection

ECHO
PING
QUIT
SELECT

Key

DEL
EXISTS
KEYS
RENAME
TYPE
EXPIRE
PERSIST
TTL
PTTL

String

APPEND
DECRBY
DECR
GET
GETSET
INCRBY
INCR
MGET
MSET
MSETNX
SET (with NX, PX, NX and XX options)
SETEX
SETNX
STRLEN

Hash

HDEL
HEXISTS
HGETALL
HGET
HKEYS
HLEN
HMGET
HMSET
HSET
HVALS

List

LPOP
LPUSH
LINDEX
LLEN
LRANGE
LSET
RPOP
RPUSH

Set

SADD
SCARD
SDIFF
SINTER
SISMEMBER
SMEMBERS
SPOP
SRANDMEMBER
SREM
SUNION

Sorted Set

ZADD
ZCARD
ZRANGEBYSCORE
ZRANGE
ZREM
ZREVRANGE
ZINCRBY

Pub/Sub

SUBSCRIBE
UNSUBSCRIBE
PSUBSCRIBE
PUNSUBSCRIBE
PUBLISH

Transactions

MULTI
EXEC
DISCARD

Scripting

EVAL
EVALSHA
SCRIPT LOAD
SCRIPT EXISTS
SCRIPT FLUSH

Design

ClauDB is implemented using Java8. Is single thread, like REDIS. It uses asynchronous IO (netty) and reactive programing paradigm (rxjava).

Requests come from IO threads and enqueues to rxjava single thread scheduler. Then IO thread is free to process another request. When request is done, the response is sended to client asyncronously. Then, every request is managed one by one, in a single thread, so there's no concurrency issues to care about.

Features

Now only implements a subset of REDIS commands, but is usable.

ClauDB also supports persistence compatible with REDIS, RDB dumps and AOF journal. It can create compatible RDB files you can load in a REDIS server.

Now ClauDB support master/slave replication, a master can have multiple slaves, but at the moment slaves can't have slaves.

Also implements partially the Pub/Sub subsystem.

Performance

Performance is quite good, not as good as REDIS, but it's good enough for Java.

This is ClauDB

$ redis-benchmark -t set,get -h localhost -p 7081 -n 100000 -q
SET: 47664.44 requests per second
GET: 50226.02 requests per second

And this is REDIS

$ redis-benchmark -t set,get -h localhost -p 6379 -n 100000 -q
SET: 97656.24 requests per second
GET: 98716.68 requests per second

In my laptop (intel core i5, with 4G of RAM)

In the latest version, ClauDB includes an option to use an off heap memory cache. See usage section

BUILD

You need to clon the repo:

$ git clone https://github.com/tonivade/claudb.git

ClauDB uses Gradle as building tool, but you don't need Gradle installed, just type:

$ ./gradlew build

This scripts automatically download Gradle and then runs the tasks.

Or if you have Gradle installed, just type

$ gradle build

Create all-in-one jar

$ gradle fatJar

DOCKER

You can create your own docker images for ClauDB using the provided Dockerfile

$ docker build -t claudb .

And then run the image

$ docker run -p 7081:7081 claudb

USAGE

You can start a new server listening in default port 7081.

$ wget http://repo1.maven.org/maven2/com/github/tonivade/claudb/1.7.1/claudb-1.7.1-all.jar
$ java -jar claudb-1.7.1-all.jar

or using jrun utility

$ jrun com.github.tonivade:claudb:com.github.tonivade.claudb.Server

Parameters:

Option        Description
------        -----------
--help        print help
-V            enable verbose log
-P            enable persistence (experimental)
-O            enable off heap memory (experimental)
-N            enable keyspace notifications (experimental)
-h <String>   host (default: localhost)
-p <Integer>  port (default: 7081)

Also you can use inside your project using Maven

<dependency>
    <groupId>com.github.tonivade</groupId>
    <artifactId>claudb</artifactId>
    <version>1.7.1</version>
</dependency>

Or gradle

compile 'com.github.tonivade:claudb:1.7.1'

Or embed in your source code

    RespServer server = ClauDB.builder().host("localhost").port(7081).build();
    server.start();

Native Image

Now is possible to generate a native image thanks to graalvm. You can generate one with this command:

$ ./gradlew clean nativeImage

Some features are not available like lua runtime and offheap memory.

TODO

Ziplist and Maplist encoding not implemented yet.
Master/Slave replication improvements. Slave with Slaves
Partitioning?
Clustering?
Geo Commands

Continuous Integration

LICENSE

ClauDB is released under MIT License

JackmanGo / claudb