Albert26193/bamboo

Bamboo

❗ the project is still under development, and the current version is 0.0.0

intro

• Bamboo is a simple, high-performance, and reliable open-source DB engine for key-value storage.
• Bamboo is written in Go, based on the Bitcask model, which supports Riak-KV
• Bamboo symbolizes rapid unidirectional growth, which is similar to the characteristics of a log-structured database.

Quick Start

• Below is a simple example of using Bamboo to store data.

package main

import (
	bamboo "bamboo/db"
	"fmt"
)

func main() {
    opts := bamboo.DefaultOptions
    opts.DataDir = "/tmp/bamboo-demo"
    db, err := bamboo.CreateDB(opts)
    if err != nil {
      panic(err)
    }

    err = db.Put([]byte("name"), []byte("bamboo"))
    if err != nil {
      panic(err)
    }
    val, err := db.Get([]byte("name"))
    if err != nil {
      panic(err)
    }
    fmt.Println("val = ", string(val))
}

Features

• Only One-Time Disk IO: Whether it is GET, PUT, or DELETE, this storage engine has at most one disk IO operation, thus ensuring high speed and predictable latency.
• Multiple Indexer: Users can use different indexers to store data, such as art-tree and btree.
• Quick Rebuild: Use MMap to quickly rebuild the index when the database is opened.

Benchmark

• the benchmark is based on the Btree indexer, and the test environment is a E5-2696v4 CPU, 512G memory, and 10T HDD.

Operation	Time (ns/op)	Memory (B/op)	Allocs (allocs/op)
Write	50423	4673	11
Read	1231	72	3
Delete	853	72	3

Future

• Support for Multiple Data Types: In the future, Bamboo will support more data types, such as list, set, and hash.
• More Benchmark: In the future, Bamboo will support more benchmark tests, such as YCSB.
• Disk IO Optimization: In the future, Bamboo will support more disk IO optimization, such as WAL.

Thanks

• Great thanks to the RoseDB, and miniDB, which provide a lot of inspiration for this project.