Redix v5

redix is a very simple key => value storage engine that speaks redis and even more simpler and flexible.

Why did I build this?

redis is very simple, sometimes we abuse it, so I decided to build a pure key-value storage system that introduces the core utilities for building any data structure you want based on the key => value model that is because I think that everything could be modeled easily using that model, so I decided to not to follow redis and all of its commands, you won't find lpush, hset, sadd, ... etc you will find a new way to do the same job but more easier and flexable, i.e, the well-known hset key field value command could be replaced with set key/field value, but sometimes you need to return a specific hashmap as key => value, but you run hget key field to get the key's value and also it could be replaced with get key/field, but how could we replace hgetall key? I will say "it is easy", let's make the hget command work as a prefix scanner that scan the whole database using the specified prefix and return all key => value pairs as redis hashmap response!, so HGETALL in redix means scan and return the result as hashmap

Features

• A really simple key => value store that speaks redis protocol but with our rules!.
• A real system that you can abuse! it isn't intedented for cache only but a "database system".
• Async (all writes happen in the background), or Sync it won't respond to the client before writing to the internal datastore.
• Pluggable storage engines, currently it supports (postgresql, filesystem), and there may be more engines be introduced in the upcomning releases.
• It could be used via redis clients easily, i.e: "the famous redis-cli"

Core Commands

• PING
• QUIT
• FLUSHALL
• FLUSHDB
• SELECT <DB index>
• SET <key> <value> [EX seconds | KEEPTTL] [NX]
• TTL <key> (not supported while using filesystem engine)
• GET <key> [DELETE], it has an alias for backward compatibility reasons called GETDEL <key>
• INCR <key> [<delta>], it has an alias for backward compatibility reasons called INCRBY (not supported while using filesystem engine)
• DEL key [key ...]
• HGETALL <prefix>

  Fetches the whole data under the specified prefix as a hashmap result

      $ 127.0.0.1:6380> set /users/u1 USER_1
      OK
  
      $ 127.0.0.1:6380> set /users/u2 USER_2
      OK
  
      $ 127.0.0.1:6380> set /users/u3 USER_3
      OK
  
      $ 127.0.0.1:6380> hgetall /users/
      1) "u1"
      2) "USER_1"
      3) "u2"
      4) "USER_2"
      5) "u3"
      6) "USER_3"
      ## in the hgetall response, redix removed the prefix you specified `/users/`

• PUBLISH <channel|topic|anyword> <message here> (not supported while using filesystem engine)
• SUBSCRIBE <channel|topic|anyword> (not supported while using filesystem engine)

Configurations

redix uses HCL for configurations, it is a very simple configuration language, Note that redix expands the environment vars in the config file, so you can use any os env var easily by wrapping it insize ${}, i.e: ${LISTEN_ADDR}

// this is the server block
// we may have multiple types of servers in future
// for now we have "redis" only
server {
    redis {
        // the listening address in the format "[hostname]:portnumber"
        listen = ":6380"

        // max connections to the same server
        // 0 means unlimited
        max_connections = 100

        // let redix tries to be async as possible as it can while writing data
        // this means that the writes will be in background which means redix won't be able to tell
        // you whether the write succeeded or not! just use this if you will use it as if it were a log-file/streaming-service
        // this config will result in a very fast writes response as redix won't block until it make sure that the writes are 100% commited!
        async = false
    }
}

// here we select the storage engine
engine "postgresql" {
    // here we define the required information to connect to the underlying datastore
    // you can pass use an env var here too i.e `dsn = "${PG_URI}"`
    dsn = "postgresql://postgres@localhost/redix"
}

Usage

• Assuming you downloaded the official docker image via docker pull ghcr.io/alash3al/redix
• Assuming that you have a configurations file called ./redix.hcl
• Just execute docker run -v $(pwd)/redix.hcl:/etc/redix/redix.hcl -p 6380:6380 ghcr.io/alash3al/redix
• Assuming that the redix-server is listening on localhost on port 6380
• Connect to it using any redis client i.e $ redis-cli -p 6380
• Nothing else!

TODOs

• Batch Write API via hmset?.
• Embedded Engine Like boltdb.

Contributions

You're welcome!