php 實現一致性hash 算法 memcache

 

散列表的應用php

涉及到數據查找比對,首先考慮到使用HashSet。HashSet最大的好處就是實現查找時間複雜度爲O(1)。使用HashSet須要解決一個重要問題:衝突問題。對比研究了網上一些字符串哈希函數,發現幾乎全部的流行的HashMap都採用了DJB Hash Function,俗稱「Times33」算法。Times33的算法很簡單,就是對字符串逐字符迭代乘以33,見下面算法原型:hash(i)=33*hash(i-1)+str[i]node

使用HashSet須要解決一個重要問題:衝突問題。對比研究了網上一些字符串哈希函數,發現幾乎全部的流行的HashMap都採用了DJB Hash Function,俗稱「Times33」算法。Times33的算法很簡單,就是對字符串逐字符迭代乘以33,見下面算法原型:hash(i)=33*hash(i-1)+str[i]算法

 

<?php
// +----------------------------------------------------------------------
// | Perfect Is Shit
// +----------------------------------------------------------------------
// | PHP實現:一致性HASH算法
// +----------------------------------------------------------------------
// | Author: alexander <gt199899@gmail.com>
// +----------------------------------------------------------------------
// | Datetime: 2017-01-11 16:01:36
// +----------------------------------------------------------------------
// | Copyright: Perfect Is Shit
// +----------------------------------------------------------------------

class ConsistentHashing
{
    // 圓環
    // hash -> 節點
    private $_ring = array();
    // 全部節點
    // 節點 -> hash
    public $nodes = array();
    // 每一個節點的虛擬節點
    public $virtual = 64;

    /**
     * 構造
     * @param array $nodes 初始化的節點列表
     */
    public function __construct($nodes = array())
    {
        if (!empty($nodes)) {
            foreach ($nodes as $value) {
                $this->addNode($value);
            }
        }
    }

    /**
     * 獲取圓環內容
     * @return array $this->_ring
     */
    public function getRing()
    {
        return $this->_ring;
    }

    /**
     * time33 函數
     * @param  string $str
     * @return 32位正整數
     * @author 大神們
     */
    public function time33($str)
    {
        // hash(i) = hash(i-1) * 33 + str[i]
        // $hash = 5381; ## 將hash設置爲0,居然比設置爲5381分佈效果更好!!!
        $hash = 0;
        //
        $s = md5($str); //相比其它版本,進行了md5加密
        $seed = 5;
        $len = 32;//加密後長度32
        for ($i = 0; $i < $len; $i++) {
            // (hash << 5) + hash 至關於 hash * 33
            //$hash = sprintf("%u", $hash * 33) + ord($s{$i});
            //$hash = ($hash * 33 + ord($s{$i})) & 0x7FFFFFFF;
            $hash = ($hash << $seed) + $hash + ord($s{$i});
        }
        return $hash & 0x7FFFFFFF;
    }

    /**
     * 增長節點
     * @param string $node 節點名稱
     * @return object $this
     */
    public function addNode($node)
    {
        if (in_array($node, array_keys($this->nodes))) {
            return;
        }
        for ($i = 1; $i <= $this->virtual; $i++) {
            $key = $this->time33($node . '-' . $i);
            $this->_ring[$key] = $node;
            $this->nodes[$node][] = $key;
        }
        ksort($this->_ring, SORT_NUMERIC);
        return $this;
    }

    /**
     * 獲取字符串的HASH在圓環上面映射到的節點
     * @param  string $key
     * @return string $node
     */
    public function getNode($key)
    {
        $node = current($this->_ring);
        $hash = $this->time33($key);
        foreach ($this->_ring as $key => $value) {
            if ($hash <= $key) {
                $node = $value;
                break;
            }
        }
        return $node;
    }

    /**
     * 獲取映射到特定節點的KEY
     * 此方法需手動調用,非特殊狀況不建議程序中使用此方法
     * @param  string $node
     * @param  string $keyPre
     * @return mixed
     */
    public function getKey($node, $keyPre = "")
    {
        if (!in_array($node, array_keys($this->nodes))) {
            return false;
        }
        $result = false;
        for ($i = 1; $i <= 10000; $i++) {
            $key = $keyPre . md5(rand(1000, 9999));
            if ($this->getNode($key) == $node) {
                $result = true;
                break;
            }
        }
        return $result ? $key : false;
    }

}

$ch_obj = new ConsistentHashing();
$ch_obj->addNode('node_1');
$ch_obj->addNode('node_2');
$ch_obj->addNode('node_3');
$ch_obj->addNode('node_4');
$ch_obj->addNode('node_5');
$ch_obj->addNode('node_6');

// +----------------------------------------------------------------------
// | 查看key映射到的節點
// +----------------------------------------------------------------------
$key1 = "asofiwjamfdalksjfkasasdflasfja";
$key2 = "jaksldfjlasfjsdjfioafaslkjflsadkjfl";
$key3 = "asjldflkjasfsdjflkajkldsjfksajdlflajs";
$key4 = "iowanfasijfmasdnfoas";
$key5 = "pqkisndfhoalnfiewlkl";
$key6 = "qjklasjdifoajfalsjflsa";
echo sprintf("%-50s 映射到節點 %s\n", $key1, $ch_obj->getNode($key1));
echo sprintf("%-50s 映射到節點 %s\n", $key2, $ch_obj->getNode($key2));
echo sprintf("%-50s 映射到節點 %s\n", $key3, $ch_obj->getNode($key3));
echo sprintf("%-50s 映射到節點 %s\n", $key4, $ch_obj->getNode($key4));
echo sprintf("%-50s 映射到節點 %s\n", $key5, $ch_obj->getNode($key5));
echo sprintf("%-50s 映射到節點 %s\n", $key6, $ch_obj->getNode($key6));

// +----------------------------------------------------------------------
// | 查看圓環和節點信息
// +----------------------------------------------------------------------
echo '<pre>';
 var_dump($ch_obj->getRing());
 var_dump($ch_obj->nodes);

// +----------------------------------------------------------------------
// | 獲取特定節點的KEY
// +----------------------------------------------------------------------
// $key1 = $ch_obj->getKey('node_1', 'pre_');
// var_dump($key1);

// +----------------------------------------------------------------------
// | 測試分佈
// +----------------------------------------------------------------------
// $keys = array();
// $rings = array();
// for ($i = 1; $i <= 60000; $i++) {
//     $key = sha1(rand(1000000,9999999));
//     $node = $ch_obj->getNode($key);
//     $rings[$node] = isset($rings[$node]) ? ++$rings[$node] : 1;
// }
// var_dump($rings);
相關文章
相關標籤/搜索