我之前的文章中说过,大多数网站的性能瓶颈不在PHP服务器上,因为它可以简单地通过横向增加服务器或CPU核数来轻松应对(对于各种云主机,增加VPS或CPU核数就更方便了,直接以备份镜像增加VPS,连操作系统、环境都不用安装配置),而是在于MySQL数据库。
如果用 MySQL 数据库,一条联合查询的SQL,也许就可以处理完业务逻辑,但是,遇到大量并发请求,就歇菜了。
如果用 NoSQL 数据库,也许需要十次查询,才能处理完同样地业务逻辑,但每次查询都比 MySQL 要快,十次循环NoSQL查询也许比一次MySQL联合查询更快,应对几万次/秒的查询完全没问题。
如果加上PHP多线程,通过十个线程同时查询NoSQL,返回结果汇总输出,速度就要更快了。我们实际的APP产品中,调用一个通过用户喜好实时推荐商品的PHP接口,PHP需要对BigSea NoSQL数据库发起500~1000次查询,来实时算出用户的个性喜好商品数据,PHP多线程的作用非常明显。
PHP扩展下载:https://github.com/krakjoe/pthreads
PHP手册文档:http://php.net/manual/zh/book.pthreads.php
1、扩展的编译安装(Linux),编译参数 --enable-maintainer-zts 是必选项:
cd /Data/tgz/php-5.5.1./configure --prefix=/Data/apps/php --with-config-file-path=/Data/apps/php/etc --with-mysql=/Data/apps/mysql --with-mysqli=/Data/apps/mysql/bin/mysql_config --with-iconv-dir --with-freetype-dir=/Data/apps/libs --with-jpeg-dir=/Data/apps/libs --with-png-dir=/Data/apps/libs --with-zlib --with-libxml-dir=/usr --enable-xml --disable-rpath --enable-bcmath --enable-shmop --enable-sysvsem --enable-inline-optimization --with-curl --enable-mbregex --enable-fpm --enable-mbstring --with-mcrypt=/Data/apps/libs --with-gd --enable-gd-native-ttf --with-openssl --with-mhash --enable-pcntl --enable-sockets --with-xmlrpc --enable-zip --enable-soap --enable-opcache --with-pdo-mysql --enable-maintainer-ztsmake cleanmakemake install unzip pthreads-master.zipcd pthreads-master/Data/apps/php/bin/phpize./configure --with-php-config=/Data/apps/php/bin/php-configmakemake install
php.ini中添加:
vi /Data/apps/php/etc/php.iniextension = "pthreads.so"
给出一段PHP多线程、与For循环,抓取百度搜索页面的PHP代码示例:
<?php class test_thread_run extends Thread { public $url; public $data; public function __construct($url) { $this->url = $url; } public function run() { if(($url = $this->url)) { $this->data = model_http_curl_get($url); } } } function model_thread_result_get($urls_array) { foreach ($urls_array as $key => $value) { $thread_array[$key] = new test_thread_run($value["url"]); $thread_array[$key]->start(); } foreach ($thread_array as $thread_array_key => $thread_array_value) { while($thread_array[$thread_array_key]->isRunning()) { usleep(10); } if($thread_array[$thread_array_key]->join()) { $variable_data[$thread_array_key] = $thread_array[$thread_array_key]->data; } } return $variable_data; } function model_http_curl_get($url,$userAgent="") { $userAgent = $userAgent ? $userAgent : 'Mozilla/4.0 (compatible; MSIE 7.0; Windows NT 5.2)'; $curl = curl_init(); curl_setopt($curl, CURLOPT_URL, $url); curl_setopt($curl, CURLOPT_RETURNTRANSFER, 1); curl_setopt($curl, CURLOPT_TIMEOUT, 5); curl_setopt($curl, CURLOPT_USERAGENT, $userAgent); $result = curl_exec($curl); curl_close($curl); return $result; } for ($i=0; $i < 100; $i++) { $urls_array[] = array("name" => "baidu", "url" => "http://www.baidu.com/s?wd=".mt_rand(10000,20000)); } $t = microtime(true); $result = model_thread_result_get($urls_array); $e = microtime(true); echo "多线程:".($e-$t).""; $t = microtime(true); foreach ($urls_array as $key => $value) { $result_new[$key] = model_http_curl_get($value["url"]); } $e = microtime(true); echo "For循环:".($e-$t)."";?>