我之前的文章中說過,大多數網站的性能瓶頸不在PHP服務器上,因為它可以簡單地通過橫向增加服務器或CPU核數來輕松應對(對於各種雲主機,增加VPS或CPU核數就更方便了,直接以備份鏡像增加VPS,連操作系統、環境都不用安裝配置),而是在於MySQL數據庫。
如果用 MySQL 數據庫,一條聯合查詢的SQL,也許就可以處理完業務邏輯,但是,遇到大量並發請求,就歇菜了。
如果用 NoSQL 數據庫,也許需要十次查詢,才能處理完同樣地業務邏輯,但每次查詢都比 MySQL 要快,十次循環NoSQL查詢也許比一次MySQL聯合查詢更快,應對幾萬次/秒的查詢完全沒問題。
如果加上PHP多線程,通過十個線程同時查詢NoSQL,返回結果匯總輸出,速度就要更快了。我們實際的APP產品中,調用一個通過用戶喜好實時推薦商品的PHP接口,PHP需要對BigSea NoSQL數據庫發起500~1000次查詢,來實時算出用戶的個性喜好商品數據,PHP多線程的作用非常明顯。
PHP擴展下載:https://github.com/krakjoe/pthreads
PHP手冊文檔:http://php.net/manual/zh/book.pthreads.php
1、擴展的編譯安裝(Linux),編譯參數 --enable-maintainer-zts 是必選項:
復制代碼 代碼如下:
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-zts
make clean
make
make install
unzip pthreads-master.zip
cd pthreads-master
/Data/apps/php/bin/phpize
./configure --with-php-config=/Data/apps/php/bin/php-config
make
make install
php.ini中添加:
復制代碼 代碼如下:
vi /Data/apps/php/etc/php.ini
extension = "pthreads.so"
給出一段PHP多線程、與For循環,抓取百度搜索頁面的PHP代碼示例:
復制代碼 代碼如下:
<?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);
$e = microtime(true);
echo "多線程:".($e-$t)."
";
$t = microtime(true);
foreach ($urls_array as $key => $value)
{
$result_new[$key] = model_http_curl_get($value["url"]);
}
$e = microtime(true);
echo "For循環:".($e-$t)."
";
?>