在大表上优化计数SQL查询

Question

I have a table with over 10 thousand registers right now, and they start to run so slow.

I have the following code:

COUNT

$SqlCount = "SELECT tabnews.New_Id
            FROM tabnew WHERE New_Id <> '' AND New_Status = 1";
$QueryCount = mysql_query($SqlCount, $Conn) or die(mysql_error($Conn));
$NumCount = mysql_num_rows($QueryCount);
$recordCount = $NumCount;

PAGINATION

if (!$id) $p = 1;
else $p = $id;
$pageSize = 16;
$itemIni = ($pageSize*$p)-$pageSize;
$totalPage = ceil($recordCount/$pageSize);

SHOW

       $Sql52 = "SELECT New_Id, New_Nome, New_Data, New_Imagem FROM tabnews WHERE New_Status = 1 ORDER BY New_Id DESC LIMIT $itemIni, $pageSize ";
       $Query52 = mysql_query($Sql52, $Conn);
       while($Rs52 = mysql_fetch_array($Query52)){

      // ECHO RESULTS
}

MY DATABASE:

CREATE TABLE IF NOT EXISTS `tabnews` (
  `New_Id` int(11) NOT NULL AUTO_INCREMENT,
  `Franquia_Id` text NOT NULL,
  `New_Slide` int(2) NOT NULL,
  `Categoria_Id` int(2) NOT NULL,
  `New_Nome` varchar(255) NOT NULL,
  `New_Data` date NOT NULL,
  `New_Imagem` varchar(75) NOT NULL,
  `New_Status` int(11) NOT NULL,
  PRIMARY KEY (`New_Id`),
  KEY `idx_1` (`New_Status`)
) ENGINE=MyISAM  DEFAULT CHARSET=latin1 AUTO_INCREMENT=10490 ;

Any ideas on how I can make this run faster?

I have a dedicated server running CENTOS.

Answer 1

20 seconds is very weird for such a little table. I have a very similar table with almost 4 million rows and your both SQL statements takes less than 0.002 sec.

CREATE TABLE IF NOT EXISTS `tasks` (
  `id` int(11) NOT NULL AUTO_INCREMENT,
  `status` varchar(10) COLLATE utf8_unicode_ci NOT NULL DEFAULT 'open',
  `method` varchar(10) COLLATE utf8_unicode_ci NOT NULL DEFAULT 'GET',
  `url` varchar(255) COLLATE utf8_unicode_ci NOT NULL,
  `params` text COLLATE utf8_unicode_ci,
  `response` text COLLATE utf8_unicode_ci,
  `executed_by` varchar(50) COLLATE utf8_unicode_ci DEFAULT '',
  `execute_at` datetime DEFAULT NULL,
  `created` datetime NOT NULL,
  `modified` datetime NOT NULL,
  PRIMARY KEY (`id`),
  KEY `status` (`status`),
  KEY `modified` (`modified`)
) ENGINE=InnoDB  DEFAULT CHARSET=utf8 COLLATE=utf8_unicode_ci AUTO_INCREMENT=3839270 ;

-

SELECT COUNT(id) FROM tasks WHERE status='done';
---> Query took 0.0008 sec

-

SELECT id, status, method, url FROM tasks WHERE status='done' ORDER BY id DESC LIMIT 200, 100;
---> Query took 0.0011 sec

Observations:

• You should use SELECT COUNT(New_Id)...
• New_id <> '' doesn't make sense. New_id can't be empty or NULL
• Set the length of New_Status to something that match the values you store there
• Try turning off logging: SET GLOBAL general_log = 'OFF';
• Update your server packages (specially MySQL)
• Is it a dedicated server only for the database?
• Is the server running other things? (run 'top' and 'uptime' to check it status)

Answer 2

This:

New_Id <> '' 

What does this do? It casts every single one of your INT primary key to string to compare it to a string. Why would you compare it to a string? It cannot be '' by definition, omit that New_Id <> '' from your WHERE clause.