An open API service indexing awesome lists of open source software.

https://github.com/huton-io/huton

An embeddable and distributed in-memory key-value store written in Go
https://github.com/huton-io/huton

cache datagrid golang

Last synced: 3 months ago
JSON representation

An embeddable and distributed in-memory key-value store written in Go

Awesome Lists containing this project

README

          

# huton
[![Go Report Card](https://goreportcard.com/badge/github.com/huton-io/huton)](https://goreportcard.com/report/github.com/huton-io/huton) [![Build Status](https://travis-ci.org/huton-io/huton.svg?branch=master)](https://travis-ci.org/huton-io/huton)

Huton is an embeddable and distributed in-memory key-value store inspired by the likes of [Redis](https://redis.io/) and [memcached](https://memcached.org/) and written in the Go programming language.

Replication is handled via the [Raft](https://raft.github.io/) consensus algorithm and data is replicated to all members of the cluster. Existing cluster members will see near-immediate consistency, while new members coming online will see eventual consistency.

Huton is considered **prototype** status, and is not yet ready for production use. It is under active development and welcomes contributions.

# Example Usage
Huton currently expects a quorum of nodes before the cluster is considered stable. This means that you need _at least_ 3 nodes to complete a cluster.
To begin, start two huton agents:
```bash
$ huton agent -name agent1 -bindPort 8100 -peers 127.0.0.1:8100,127.0.0.1:8200 -expect 3
$ huton agent -name agent2 -bindPort 8200 -peers 127.0.0.1:8100,127.0.0.1:8200 -expect 3
```
Huton requires that the name of each node in a cluster be unique. If one or more nodes have the same name, this can cause havoc with the internal peer list.

It should also be noted that even though in this example, we are seeding the cluster with huton agents, the agents aren't really needed at all. A cluster can (and probably will, in a real world scenario) consist entirely of huton instances embedded in an application.

Once the two agents are started, only one more node is needed to reach quorum. If you run the below Go code, it will join the cluster and provide an HTTP resource to get and put data.
```go
package main

import (
"strings"
"net/http"
"github.com/huton-io/huton/lib"
"flag"
)

func main() {
var instanceName string
var peers string
var port int
flag.StringVar(&instanceName, "name", "", "unique instance name")
flag.StringVar(&peers, "peers", "", "comma delimited list of seed peers")
flag.IntVar(&port, "port", 8300, "bind port")
flag.Parse()
if instanceName == "" {
panic("No instance name provided")
}
if peers == "" {
panic("No peers provided.")
}
instance, err := huton.NewInstance(instanceName, huton.BindPort(port), huton.BootstrapExpect(3))
if err != nil {
panic(err)
}
_, err = instance.Join(strings.Split(peers, ","))
if err != nil {
panic(err)
}
http.HandleFunc("/", handler(instance))

go func () {
if err = http.ListenAndServe("127.0.0.1:8080", nil); err != nil {
panic(err)
}
}()
signalCh := make(chan os.Signal, 1)
signal.Notify(signalCh, os.Interrupt, syscall.SIGTERM, syscall.SIGKILL)
<-signalCh
c.instance.Shutdown()
}

func handler(instance huton.Instance) func(w http.ResponseWriter, req *http.Request) {
return func(w http.ResponseWriter, req *http.Request) {
cache, err := instance.Cache("exampleBucket")
if err != nil {
http.Error(w, "Failed to get bucket.", http.StatusInternalServerError)
return
}
switch req.Method {
case http.MethodGet:
ss := cache.Snapshot()
val := ss.Get([]byte("key"))
w.Write(val)
case http.MethodPost:
batch := cache.NewBatch(1, 8)
if err = batch.Set([]byte("key"), []byte("value")); err != nil {
http.Error(w, err.Error(), http.StatusInternalServerError)
return
}
if err = cache.ExecuteBatch(batch); err != nil {
http.Error(w, err.Error(), http.StatusInternalServerError)
}
}
}
}

```
The above code example illustrates how you can embed Huton within an application.

Start the above program:
```bash
go run main.go -name app -bindPort 8300 -peers 127.0.0.1:8100,127.0.0.1:8200
```

Once all three nodes are started and connected to each other, put some data:
```bash
curl -v -X POST http://127.0.0.1:8080/
```
Now retrieve the data:
```bash
curl -v http://127.0.0.1:8080/
```
The HTTP resource should return the text _value_.