Splitting the sql-file
I just read every line in with PHP and when I encounter a comment (– ), I create a new file.

//Set time limit to one hour
set_time_limit(3600);

$dir = '/path/to/store/splits';
$file = '/path/to/large/sql/file.sql';

$i = 1000000;
//open large sql-file
$handle = fopen($file,"r");
if($handle) {
	while(($buffer = fgets($handle)) !== false) {
		//read line and append it to the file with name $i.sql
		$newfile = fopen($dir . '/' .$i . '.sql', 'a+');
		fwrite($newfile,$buffer);
		//if a comment is found, create a new file
		if(substr($buffer,0,3) === '-- '){
			$i++;
		}
	}
fclose($handle);
}

Splitting this 20Gb file took about 10 minutes on my server and created approximately 500 files. The largest file was 5Gb and I know I could import a 5Gb-file without a problem.

Importing the sql-file
Importing the file was pretty simple. I wrote a script that loops over all the files and imports them into the database.

set_time_limit(3600000);

//The directory containing the splitted files
$dir = '/path/to/store/splits';

$files = scandir($dir);

ob_start();
//loop over the files and import them into the database
foreach($files as $file){
	echo $dir . $file . "\n";
	system("mysql -u mysql_username -pmysql_password mysql_databasename < $dir/$file");
	ob_flush();
}

In my case, it took more than 8 hours to completely import those database but with the help of these 2 scripts, it worked like a charm.

For this test, I’ve set up a server (1 intel processor/512MB ram/12MB video memory) with Ubuntu Server 10.10 (32 bit).
I could test some static resources but nowadays, particularly all websites are dynamic ones so I’ve installed Mysql/PHP and of course Apache2, all with the default settings, on my server.
Next, I’ve downloaded WordPress and installed it on the server, changed the appearance to ‘pixel theme’ and added a few posts with images in.

Time to test

I’m not going to do fancy stuff, the module is installed with the default options enabled and I’ll just use Firebug to do the tests.

First the tests without the module installed:

The results with the module installed:

Mod_pagespeed_statistics:

Conclusion:

• The amount of requests have been reduced in this case with 2 (-15%)
• The pagesize has been reduced with 5.2KB (-1.5%)
• The loadtime is shorter when mod_pagespeed is enabled (-13%)

The 50% profit claim is a little bit optimistic because the bottleneck in the pageload is the dynamic content and that’s a part where this module doesn’t do anything. But in general, the module does a reasonably good job. I get a 13% profit with mostly default values, a good configured server should propably even do better.
I will definitely test this further and maybe put it one day on a production server.

The config used in this test looks like this:

<IfModule pagespeed_module>
    SetOutputFilter MOD_PAGESPEED_OUTPUT_FILTER
     ModPagespeed on
     ModPagespeedUrlPrefix "http://localhost/"
     ModPagespeedFileCachePath "/var/mod_pagespeed/cache/"
     ModPagespeedGeneratedFilePrefix "/var/mod_pagespeed/files/"
     ModPagespeedRewriteLevel CoreFilters
     ModPagespeedDomain localhost
     ModPagespeedEnableFilters add_instrumentation
<IfModule>

My full configuration file can be downloaded here.

package be.eek.test.mysql;

import java.io.IOException;
import java.io.InputStream;

public class MysqlBackup {

    /**
    main method is just to create a working example
    **/
    public static void main(String[] args) throws IOException {
        new MysqlBackup().MakeBackup();
    }

    public void MakeBackup() throws IOException {
        String dump = "mysqldump "      //Path to mysql
                + "--host=localhost "       //Mysql hostname
                + "--port=3306 "            //Mysql portnumber
                + "--user=root "            //Mysql username
                + "--password=test "        //Mysql password
                + "--add-drop-table "       //Add a DROP TABLE statement before each CREATE TABLE statement
                + "--add-drop-database "    //Add a DROP DATABASE statement before each CREATE DATABASE statement
                + "--complete-insert "      //Use complete INSERT statements that include column names.
                + "--extended-insert "      //Use multiple-row INSERT syntax that include several VALUES lists
                + "test";                   //Mysql databasename

        Process run = Runtime.getRuntime().exec(dump);

        InputStream in = run.getInputStream();

        int nextChar;
        StringBuffer sb = new StringBuffer();

        while ((nextChar = in.read()) != -1) {
            sb.append((char) nextChar);
        }

        //Here, you can for example write it to a file and save it
        System.out.println(sb);
    }
}

The same thing is also possible in php:

$dump = "mysqldump "
        . "--host=localhost "       //Mysql hostname
        . "--port=3306 "            //Mysql portnumber
        . "--user=root "            //Mysql username
        . "--password=test "        //Mysql password
        . "--add-drop-table "       //Add a DROP TABLE statement before each CREATE TABLE statement
        . "--add-drop-database "    //Add a DROP DATABASE statement before each CREATE DATABASE statement
        . "--complete-insert "      //Use complete INSERT statements that include column names.
        . "--extended-insert "      //Use multiple-row INSERT syntax that include several VALUES lists
        . "test";                   //databasename

$backup = system($dump);
echo $backup;
die();