如何加快MySQL查询？ 从3个表格加载数据

Question

I have 3 tables. First: "atributy" Second: "atributy_value" Third: "produkty"

I have this query first:

SELECT a.*, p.ATTRIBUTE_CODE, p.ATTRIBUTE_VALUE, p.KATEGORIA  
FROM atributy a JOIN produkty p ON p.ATTRIBUTE_CODE
LIKE CONCAT('%', a.code, '%')
AND
KATEGORIA IN ('$kategoria_sql') 
GROUP BY a.value

And my second query is this:

SELECT * FROM atributy_value
INNER JOIN produkty 
ON produkty.ATTRIBUTE_VALUE LIKE CONCAT('%', atributy_value.ValueCode, '%')
AND AttributeCode = '$atribut_kod' 
AND KATEGORIA IN ('$kategoria_sql')
GROUP BY atributy_value.Value

Help me please make from this 2 query's 1 one better. Reason: too long loading my web e-shop.

EDIT:

$query = mysql_query("
SELECT a.*, p.ATTRIBUTE_CODE, p.ATTRIBUTE_VALUE, p.KATEGORIA  
FROM atributy a JOIN produkty p ON p.ATTRIBUTE_CODE LIKE CONCAT('%', a.code, '%') 
AND KATEGORIA IN ('$kategoria_sql') 
GROUP BY a.value ");

while($result = mysql_fetch_object($query)){

$atribut_kod = $result->code;
$atribut_value = $result->value;
$nazov_produktu = $result->NAZOV;
$value1 = $result->ATTRIBUTE_VALUE;
$value1 = explode(" ", $value1);

$value1_count = count($value1);

echo "<div class=\"parametre_panel\">
<h3>".$atribut_value."</h3>
";

$url_kody .= "$atribut_kod,";

$hodnoty_qry = mysql_query("
SELECT * FROM atributy_value
INNER JOIN produkty ON produkty.ATTRIBUTE_VALUE LIKE CONCAT('%', atributy_value.ValueCode, '%')
AND AttributeCode = '$atribut_kod' 
AND KATEGORIA IN ('$kategoria_sql')  
GROUP BY atributy_value.Value ");

 while($hodnoty_res = mysql_fetch_object($hodnoty_qry)){
 $cislo_hodnoty = $hodnoty_res->ValueCode;
 echo "<input type=\"checkbox\" class=\"ZobrazParametrickeVyhladavanie\"   name=\"value[]\" id=\"$cislo_hodnoty\" value=\"".$atribut_kod."-".$cislo_hodnoty."\"><label for=\"$cislo_hodnoty\">".$hodnoty_res->Value."</label>
 ";

 $url_hodnoty .= "$cislo_hodnoty,";
} //second query while()
echo "</div>";
} //first query while()

EDIT 2:

My table structure

produkty:       http://i.imgur.com/J4Kz2CE.png 
atributy_value: http://i.imgur.com/nX1uRph.png 
atributy:       http://i.imgur.com/mlCa3It.png

Indexes:

atributy:         http://i.imgur.com/ppMEEOe.png
atributy_value:   http://i.imgur.com/RHAeSiu.png
produkty:         http://i.imgur.com/IUrgy9l.png

Answer 1

You can try Below Query:

SELECT a.*, 
       p.ATTRIBUTE_CODE, 
       p.ATTRIBUTE_VALUE, 
       p.KATEGORIA  
  FROM atributy a 
  JOIN (SELECT * 
          FROM atributy_value
         INNER JOIN produkty 
            ON produkty.ATTRIBUTE_VALUE LIKE CONCAT('%', atributy_value.ValueCode, '%')
           AND AttributeCode = '$atribut_kod' 
           AND KATEGORIA IN ('$kategoria_sql')
         GROUP BY atributy_value.Value) p 
    ON p.ATTRIBUTE_CODE LIKE CONCAT('%', a.code, '%')
   AND KATEGORIA IN ('$kategoria_sql') 
 GROUP BY a.value

In Inner Query Select Only Required Columns

Hope this works.

Answer 2

Two things are slow here.

First - unnecessary outer / inner loop. It should be possible to do this in a single SQL query, which will be a huge time saving. Without seeing the definitions of your tables, this is the best I can suggest :

 SELECT a.*, p.*, av.*
 FROM produkty p 
 JOIN atributy a ON p.ATTRIBUTE_CODE LIKE CONCAT('%', a.code, '%') 
 JOIN atributy_value av ON p.ATTRIBUTE_VALUE LIKE CONCAT('%', av.ValueCode, '%') 
 WHERE 
 KATEGORIA IN ('$kategoria_sql') 

At least it's a starting point to test from.

Second - Joining between the two tables using LIKE '%value%'. Is there not an integer ID that links these two tables together? Or at least, should the text be an exact match and you can remove the '%'s?

EDIT (after adding table structure) :

You have two joins, the first from a nice indexed integer atributy_value.ValueCode to a non-indexed text field produkty.ATTRIBUTE_CODE, the second from a non indexed text field atributy.code to another non-indexed text field produkty.ATTRIBUTE_CODE.

It's not a good table structure, it would be easier and faster if every table had a unique integer ID to join on. But you might not have time to change this.

Can you just remove the %'s in your original queries and insist on an exact match between the columns? To keep it simple, you could just replace :

LIKE CONCAT('%', a.code, '%')

replace with

= a.code

and

LIKE CONCAT('%', atributy_value.ValueCode, '%') 

replace with

= CONVERT( varchar(500), atributy_value.ValueCode)

This will make it faster. If it's still not enough you could add indexes on Produkty.ATTRIBUTE_CODE and Produkty.ATTRIBUTE_VALUE and atributy.code.

Answer 3

1) Run an EXPLAIN on your queries and add indexes where necessary. IMHO, these queries are not meant to run on production. However, I have made a suggestion below.

2) Let MySQL handle the query execution plan for you. Instead of mentioning a JOIN. Try this

$query = mysql_query("
                        SELECT a.*, p.ATTRIBUTE_CODE, p.ATTRIBUTE_VALUE, p.KATEGORIA 
                        FROM atributy a,produkty p 
                        WHERE p.ATTRIBUTE_CODE LIKE CONCAT('%', a.code, '%') 
                        AND KATEGORIA IN ('$kategoria_sql') 
                        GROUP BY a.value ");

2) Also change the inner query to the following:

SELECT atributy_value.* FROM atributy_value AS atributy_value, produkty AS produkty  
WHERE 
produkty.ATTRIBUTE_VALUE LIKE CONCAT('%', atributy_value.ValueCode, '%')
AND atributy_value.AttributeCode = '$atribut_kod' 
AND produkty.KATEGORIA IN ('$kategoria_sql')  
GROUP BY atributy_value.Value ")

I need more details on the table structures though

3) Run an EXPLAIN on the above query. If you have any missing indexes, please add them