A distributed, fault-tolerant key-value store.

The data is persisted on the disk using BadgerDB. Replication and fault-tolerance is achieved using Raft.

It also provides a simple HTTP API for accessing the data.

Installation

From Source

1. Clone the repository

git clone https://github.com/rohankmr414/arima.git
cd arima

2. Build the binary

go mod download
go build ./cmd/arima
sudo mv ./arima /usr/local/bin/

This will build the arima binary and place it in the /usr/local/bin directory. You can now run the binary using arima.

Usage

$ arima --help
NAME:
   arima - A simple fault tolerant key-value store

USAGE:
   arima [global options] command [command options] [arguments...]

DESCRIPTION:
   A distributed fault-tolerant key-value store which uses Raft for consensus.

COMMANDS:
   run, r   Run the server
   help, h  Shows a list of commands or help for one command

GLOBAL OPTIONS:
   --help, -h  show help (default: false)

Running a node

$ arima run --server-port 2221 --node-id n1 --raft-port 1111 --volume-dir /tmp/arima/n1

Settting up a cluster

• Run multiple nodes.

  $ arima run --server-port 2221 --node-id n1 --raft-port 1111 --volume-dir /tmp/arima/n1
  $ arima run --server-port 2222 --node-id n2 --raft-port 1112 --volume-dir /tmp/arima/n2
  $ arima run --server-port 2223 --node-id n3 --raft-port 1113 --volume-dir /tmp/arima/n3
  

  Each node will be initialized as a leader. We'll have to connect the nodes to form a cluster.

• Connect the nodes to form a cluster.

  Manually pick one node as the leader and connect it to the rest of the nodes.

  For node n2

  $ curl --location --request POST 'localhost:2221/raft/join' \
  --header 'Content-Type: application/json' \
  --data-raw '{
      "node_id": "n2", 
      "raft_address": "127.0.0.1:1112"
  }'
  

  For node n3

  $ curl --location --request POST 'localhost:2221/raft/join' \
  --header 'Content-Type: application/json' \
  --data-raw '{
      "node_id": "n3", 
      "raft_address": "127.0.0.1:1113"
  }'
  

  Then, check each of this endpoint, it will return the status that the port 2221 is now the only leader and the other is just a follower:

  • http://localhost:2221/raft/stats
  • http://localhost:2222/raft/stats
  • http://localhost:2223/raft/stats

  A 3 node cluster is now formed.

  
  

Reading and Writing Data

Once the cluster is formed, we can start sending HTTP requests to the leader node to read, write and delete key-value pairs.

We can read the data from any node in the cluster.

Each node exposes following endpoints:

• URL: /store/:key

  • Method: GET
  • Response: 200

    {
        "data": {
            "key":   "key",
            "value": "value"
        },
        "message": "success fetching data"
    }

• URL: /store/

  • Method: POST
  • Request:

    {
        "key":   "key",
        "value": "value" 
    }

  • Response: 200

    {
        "data": {
            "key":   "key",
            "value": "value" 
        },
        "message": "success fetching data"
    }

• URL: /store/:key

  • Method: DELETE
  • Response: 200

    {
        "data": {
            "key":  "key",
            "value": null
        },
        "message": "success removing data"
    }

Removing a node

• URL: /raft/remove
  • Method: POST
  • Request:

    {
        "node_id": "n1",
        "raft_address": "127.0.0.1:1111"
    }

  • Response: 200

    {
        "data":{
            ...node stats,
        },
        "message":"node n1 removed successfully"
    }