I want to be free

You might be surprised, but a right choise of the project text encoding can affect the project file size and amount of bugs.

To avoid bugs of wrong presentation of text on your page, make sure that all database entities and the application server (PHP) use the same encoding. That helps to forget about issues connected with text presentation.

On database side, make sure you set the correct encoding to:

• database,
• tables,
• columns,
• import-export tools parameters (a big source of wrong encoding bugs),
• corresponding SQL server variables

On application server it’s usually just one query –

SET NAMES utf8

Pay attention, that utf8 might be not the best choise for your project: every non-English character needs 2-6 bytes of memory, so if you built a one-language (local) project with lots of database data, consider using a 1 byte encoding like windows-1251 and save about half of the space on server file system.

A foreign key, this classic concept of relational databases, is extremely helpful in many ways.

To start with, it’s a watch dog of your database integrity — for example, it would be impossible to add user images records, if the user does not exist. No additional code, more control.

Another benefit is that all depending data is deleted cascadedly when the parent row gets killed. You remove a user — and voila: all images, comments, ratings and staff fly away too.

And third, it helps to autogenerate admin tools; for instance, in Symfony framework, the generator of admin interface relies on this info.

To be able to use foreign keys in your MySQL database, do 2 following steps:

• make sure the two tables you are connecting, are driven by InnoDB
• ensure that the columns you connect (user_id in our example) are of exactly the same type, including NULL and sign allowance.

We at SunnyRentals are working a lot with geo data.

The last task I was solving was inclusion of all cities in a touristic regions bounds automatically.

Yes, there are MySQL spatial data types, so region border and cities coordinates can be stored in native data types, but the processing functions are not implemented right: they only operate on MBRs (minimum bounding rectangles) to make things simpler…

So, let’s create our own MySQL function for that.

There are different approaches to distinguish if a point is within a given polygon. I prefered the way to iterate the edges of the polygon and check where the point relatively to the edge is – at right or at left; if the point is always at the same side, it lies within the border.

So we came to a fact that we need 2 functions:

1. Distinguish the side at which the point lies from the line
2. Check that for every edge in the polygon.

Here is the first function:

DELIMITER //

DROP FUNCTION `GetPointPositionOfLine`//
CREATE FUNCTION `GetPointPositionOfLine`( a POINT, b POINT, p POINT) RETURNS tinyint NO SQL
  COMMENT 'Distinguish the side at which the point lies from the line'
BEGIN
  DECLARE s DOUBLE;
  SET s = (X(b) - X(a))*(Y(p) - Y(a)) - (Y(b) - Y(a))*(X(p) - X(a));
  IF s > 0 THEN
    RETURN 1;
  END IF;
  IF s < 0 THEN
    RETURN -1;
  END IF;
  RETURN 0;
END //

All three params are of type POINT. Points a and b are the beginning and end of the line. Point p is the point we check.

Here is the main function:

DELIMITER //

DROP FUNCTION `IsPointWithin`//
CREATE FUNCTION `IsPointWithin`( ls LINESTRING, p POINT) RETURNS tinyint(1) NO SQL
BEGIN
  DECLARE i INT;
  DECLARE a, b POINT;
  DECLARE LineStringNumPoints INT;
  DECLARE Position, PrevPosition TINYINT;

  SET i = 1;
  SET PrevPosition = 0;
  SET LineStringNumPoints = NumPoints(ls);

  WHILE i <= LineStringNumPoints DO
    SET a = PointN( ls, i);
    IF i = LineStringNumPoints THEN
      SET b = StartPoint( ls);
    ELSE
      SET b = PointN( ls, i+1);
    END IF;

    SET Position = GetPointPositionOfLine( a, b, p);
    IF Position <> 0 THEN
      IF Position <> PrevPosition AND PrevPosition <> 0 THEN
        RETURN FALSE;
      END IF;
      SET PrevPosition = Position;
    END IF;
    SET i = i + 1;
  END WHILE;
  RETURN TRUE;
END //

It checks if a point p lies within a polygon of type LINESTRING.

Usage example. Given you have coordinates of cities in the separate float columns (to keep them readable) lat and lng of city table, and the polygon is given as a string too — but of course you can use a column of type LINESTRING:

SELECT *
FROM `city` c
WHERE IsPointWithin(GeomFromText('LINESTRING(4 8, 15 16, 23 42)'), POINT(c.`lat`, c.`lng`));

The solution works pretty fast, especially if you first limit the cities by the rectangle that the polygon lies in.

At times it’s necessary to know which queries the server runs right now.

Of course, you can try this statement:

SHOW PROCESSLIST

but if you need a more poweful tool, try MySQL Proxy.

Although it’s still an alpha version, it’s a handy tool to monitor, filter and manipulate your queries since there is Lua scripting language interpreter supported which is quite obvious in usage.

The MySQL Proxy (as can be understood from its title) can help you to see communication between a few MySQL servers and a few clients.

My usage pattern is that the Proxy listens to 4040 port on my local machine, and if I change the port setting in connection string of any application (my job project or even PhpMyAdmin), I can see the queries logged in a console window.

You can track lots of characteristics of queries — just take a look at example Lua scripts in the package that you’ve downloaded.

If you want to load a list of SQL files into your database on Windows, you can create a .cmd file with such content and run it.
Place it in the same folder where your .sql files are.

@echo off
SET DATABASE=my_database
SET USER=root
SET PASS=1
SET FORMAT=*.sql

FOR /F "usebackq" %%i IN (`dir /on /b %FORMAT%`) DO echo %%i && mysql --user=%USER% --password=%PASS% --default-character-set=utf8 %DATABASE% < %%i

Make sure, mysql command can be run. If no, either set the full path to mysql command tool, or add the path to PATH environment variable.

Sphinx is a nice search engine and is used at our project hugely.

Although, I found a case when using a pre-fetch query is not obvious. The pre-fetch query is a query that is run before the data is got from database, so thus you can set character encoding, tune database caching, set variables, etc.

So, if we have 2 sources, one is parent for all other sources since it has database connection params, and second one inherits from the parent.

source www {
    type                    = mysql

    sql_user                = root
    sql_pass                = 

    sql_query_pre           = SET NAMES utf8
}

source www_country : www {
    sql_query = SELECT * FROM country
}

All works fine – when www_country souce is used, the same sql_query_pre will be used because www_country is a child of parent source.

Although, if one day you decide to add a custom pre-query to the child source, it seems you loose the parent’s ones:

...
source www_country : www {
    sql_query_pre = SET @usefulVariable = 1

    sql_query = SELECT * FROM country
}

In this case it seems you have to duplicate the parental pre-query in every child =(
So the right child is gonna be:

...
source www_country : www {
    sql_query_pre = SET NAMES utf8
    sql_query_pre = SET @usefulVariable = 1

    sql_query = SELECT * FROM country
}

Reverse geocoding is a process of getting toponym name (city name) by its coordinates.

To make this, you can use a remote service like I have already described, or implement it yourself.

To do so you need two things:

1. A database of cities with coordinates (having a population value at least for major cities is a big plus)
2. A method to find a closest city

#1 is left for you, let’s talk about #2.

In the previous post I introduced the DISTANCE function (SQL) to count a distance between 2 points by coordinates. Let’s use it to find the biggest city closest to a given point:

DELIMITER //
DROP FUNCTION IF EXISTS REVERSE_GEOCODE;//
CREATE FUNCTION REVERSE_GEOCODE( latitude DOUBLE, longitude DOUBLE, radius DOUBLE )
    RETURNS DOUBLE READS SQL DATA DETERMINISTIC
    COMMENT 'Finds a city by coordinates given. Returns City ID.'
BEGIN
    DECLARE cityID INT(10) DEFAULT 0;
    DECLARE d DOUBLE DEFAULT 0; /* useless var */

    SELECT id, DISTANCE(`lat`, `lng`, latitude, longitude) AS dist
    INTO cityID, d
    FROM `geo_cities`
    WHERE lat BETWEEN FLOOR(latitude - 0.5)
                  AND CEIL(latitude  + 0.5)
      AND lng BETWEEN FLOOR(longitude- 0.5)
                  AND CEIL(longitude + 0.5)
    HAVING dist < radius
    ORDER BY IF(population > 0, population, 1) * (1/IF(dist > 0, dist, 1)) DESC
    LIMIT 1;

    RETURN cityID;
END; //
DELIMITER ;

Well, this function orders cities by (population and distance) function — so that the biggest closest city is returned.

WHERE clause is supposed to shorten the number of cities which are looked through; otherwise we would have to search through whole world’s cities. In my solution the search is limited by a reasonably small rectangle.

In the next post I’ll tell you how to find a biggest city depending on current Google Maps zoom, so that it won’t give you a small city near Paris if you click on France at a world-level map view.

Useful function.

DELIMITER //

DROP FUNCTION IF EXISTS DISTANCE; //

CREATE FUNCTION DISTANCE( lat1 DOUBLE, lon1 DOUBLE, lat2 DOUBLE, lon2 DOUBLE )
    RETURNS DOUBLE NO SQL DETERMINISTIC
    COMMENT 'counts distance (km) between 2 points on Earth surface'
BEGIN
    DECLARE dtor DOUBLE DEFAULT 57.295800;

    RETURN (6371 * acos(sin( lat1/dtor) * sin(lat2/dtor) +
        cos(lat1/dtor) * cos(lat2/dtor) *
        cos(lon2/dtor - lon1/dtor)));
END; //

DELIMITER ;

In next post I will explain how I used this function to implement a reverse geocoding — it’s useful when you need to find a closest city by given coordinates.

At times, some records in a recordset are more important than others — e.g. a featured product in a products list. In such a case you can make 2 requests to get the featured item first, then to load other N-1 items, and then display them.

If you want to make it by one move, you can use a feature of preferrable sorting — you can define what value and in which field goes first if such value is present.

For instance, you want to show an item with item_id=15 first:

SELECT *
FROM items
ORDER BY item_id IN (15) DESC

Which returns something like (featured item goes first):

+----------+------------+
| item_id  | name       |
+----------+------------+
| 15       | Product 15 |
| 32       | Product 32 |
| 31       | Product 31 |
| 30       | Product 30 |
| 29       | Product 29 |
| ...      | ...        |
+----------+------------+

If you import unicode text into latin1 database column, the symbols would be screwed up — russian symbols become a shit like “залоговый депозит”.

To convert such quickly (as a test) you can use Lebedev’s convertor.

To convert the whole table do the following (thanks to dull.ru):

1. The most important step: dump such data in a file with mysqldump

   mysqldump -u user -p --default-character-set=latin1 --skip-set-charset --no-create-info --extended-insert --complete-insert dbname table > dbname.sql

   If not the whole table data is bad, create another table (CREATE TABLE t2 LIKE t1) and copy wrong rows to the new table.

2. Replace latin1 by utf8 in the file
   This can be done manually in your editor or by this command:

   sed -r 's/latin1/utf8/g' dbname.sql > dbname_utf.sql

3. Convert your latin1-table to utf8 (and maybe truncate it):

   ALTER TABLE `table` CONVERT TO CHARACTER SET 'utf8';

4. Import the utf8 data back to the table:

   mysql -u user -p --default-character-set=utf8 dbname < dbname_utf.sql

That’s it.

Are you aware of Federated engine in MySQL (apart from MyISAM and InnoDB)?

This engine allows you to define a table that sucks data from another table, even from a remore server. The tables definition must be the same.

I use it for the following:

1. Every time I rebuild the project, I have wait for 15 minutes while two big tables are created and filled with data — these are geo data tables (world cities, regions, etc), 4 mln records, and POI table, 2 mln records. I use Federated tables to create two separate databases and just link these tables in my project.
2. These tables are shared between several environments (dev, test and live) on the same server.

To check if your MySQL server has the Federated engine supported, you can use just a phpMyAdmin — go to home page of you phpMyAdmin installation (click Home picture), then choose Engines tab and check there.

If it’s not enabled (gray), open your my.ini file, find the “[mysqld]” part and make it to look like this:

[mysqld]
federated

P.S. If you have an error in the table definition, phpMyAdmin shows your database as empty. To fix this, log in via mysql console and try to make a SELECT from this poorly defined table and you get the error message to work with.

That’s easy, you need 2 things:

1. Fresh currencies exchange rates
2. Some way to excange amount from one currency to another.

This how I did it: get values from European Central Bank (ECB) for step #1 and wrote MySQL user defined function for step #2.

Here is how to export currencies rates from ECB (EUR is a base currency, and I add self rate as 1:1). First I create such database table:

CREATE TABLE IF NOT EXISTS `currency` (
  `code` char(3) NOT NULL DEFAULT '',
  `rate` decimal(10,5) NOT NULL COMMENT 'Rate to EUR got from www.ecb.int',
  PRIMARY KEY (`code`)
) ENGINE=InnoDB DEFAULT CHARSET=utf8 COMMENT='Currency rates (regularly updated)';

Now let’s fill it with rates:

<?php

class CurrencyController extends Controller_Ajax_Action {

  public function importAction() {

    $db = Zend_Registry::get('db');
    $db->beginTransaction();

    $url = 'http://www.ecb.int/stats/eurofxref/eurofxref.zip?1c7a343768baab4322620e3498553b5a';
    try {
      $contents = file_get_contents($url);
      $contents = archive::unzip($contents);
      $contents = explode("\n", $contents);

      $names = explode(',', $contents[0]);
      $rates = explode(',', $contents[1]);

      $names[] = 'EUR';
      $rates[] = 1;

      for ($i = 1; $i < sizeof($names); $i++) {
        if (!(float) $rates[$i]) continue;
        $db->query( sprintf('INSERT INTO `currency`(`code`, `rate`)
              VALUES ("%s", %10.5f)
              ON DUPLICATE KEY UPDATE `rate`=VALUES(`rate`)',
             trim( $names[$i] ),
             trim( $rates[$i] )
        ) );
      }

      $db->commit();
    } catch ( Exception $O_o ) {
      error_log( $O_o->getMessage() );
      $db->rollback();
    }

  }
}

Now let’s create a SQL function for handy converts. Create a udf.sql file and add this in it:

DELIMITER //

DROP  FUNCTION IF EXISTS EXCHANGE;
CREATE FUNCTION EXCHANGE( amount DOUBLE, cFrom CHAR(3), cTo CHAR(3) ) RETURNS DOUBLE READS SQL DATA DETERMINISTIC
    COMMENT 'converts money amount from one currency to another'
BEGIN
    DECLARE rateFrom DOUBLE DEFAULT 0;
    DECLARE rateTo DOUBLE DEFAULT 0;

    SELECT `rate` INTO rateFrom FROM `currency` WHERE `code` = cFrom;
    SELECT `rate` INTO rateTo   FROM `currency` WHERE `code` = cTo;

    IF ISNULL( rateFrom ) OR ISNULL( rateTo ) THEN
        RETURN NULL;
    END IF;

    RETURN amount * rateTo / rateFrom;
END; //

DELIMITER ;

and run this command in your shell:

mysql --user=USER --password=PASS DATABASE < udf.sql

This is how you can use this function — how to convert 10 US dollars to Canadian dollars:

SELECT EXCHANGE( 10, 'USD', 'CAD')

which results in $10 = 10.93 Canadian dollars.

P.S. Consider adding the currency export action call to your cron scripts.

P.P.S. A function to unzip the data file can be got at php.net

Data feeds often come in XML format, so your application must be able to deal with that format.

As I already wrote, data in CSV format (comma separated values) can be loaded to database extremely fast. So my idea was to convert XML data files to CSV and then use bulk load to database. My tests shown that this is faster in 10-100 times than one by one inserts.

Yesterday I decided to write a generalized solution for this, and it turned out that there is no need: it’s just coming — MySQL 6 will have such feature!

How it works: you create a table, name its columns exactly as XML nodes/attributes names or — and MySQL server will load it correspondently.

Example — you downloaded a POI list file (Points of Interest) called poi.xml that looks like this:

<?xml version="1.0" encoding="UTF-8"?>
<gpx>
    <wpt lat="58.691931900" lon="11.253125962">
        <name>Parking</name>
    </wpt>
    <wpt lat="58.315525000" lon="12.305828000">
          <name>Fast Food Restaurant:Max i trollhattan</name>
    </wpt>
    <wpt lat="57.717958100" lon="11.880860600">
          <name>Picnic spot</name>
    </wpt>
</gpx>

You create a MySQL table:

CREATE TABLE IF NOT EXISTS `poi` (
  `lat` varchar(255) NOT NULL,
  `lon` varchar(255) NOT NULL,
  `name` varchar(255) NOT NULL
) DEFAULT CHARSET=utf8;

OK, now you load the XML data to your table:

LOAD XML INFILE '\\path\\to\\poi.txt'
INTO TABLE `poi`
ROWS IDENTIFIED BY '<wpt>'

Voila!

The good thing is that MySQL 6 is already available in alpha version — good enough for development purposes; I gave it a try — it takes 5 seconds to load 4.8 Mb of data in 19 files.

There are situations when you need to separate processing of big amount of data between several “agents”, e.g.:

• you have a long list of websites which must be checked for being alive (404 error check) by your web-clawlers;
• a queue of photos to be resized or videos to be converted;
• articles that your editors must review;
• catalogue of blog feeds that your system must import posts from;
• etc.

The idea to do this is simple:

1. Give a small piece of big work to an agent.
2. Mark this piece as given to him (so that none other starts to do the same job) and remember the time stamp when the job was given or when the job becomes obsolete (this agent is dead, let’s give this job to someone else).
3. If work is done — go to step #1.
4. After some period of time (1 hour) check all the time stamps, and if some agents didn’t cope with the job, mark the jobs as free so that others could start to work on it.

The problem is between steps #1 and #2 — while you gave a job to Agent 1 and going to mark it as given to him, what if Agent 2 is given by the same job? If you have many Agents, this  can happen at real. This situations is called concurrent read/write.

To overcome this a lock can be used.

In this article I wil explain, how to use locks in Zend project with MySQL database.

First of all, MySQL documentation tells that SELECT .. FOR UPDATE can be used for that purpose. First step is to select records by that statement, and second step is to mark them as locked. Requirements are to use InnoDB storage and to frame these two statements in a transaction.

Happily, Zend_Db_Table_Select has a special method forUpdate() that implements SELECT .. FOR UPDATE statement. Zend_Db can cope with transactions as well. Let’s try it!

To lock a record, we need two fields:

1. one to remember ID of agent that is processing this record (let’s call this column ‘locked_by‘)
2. one another to know the time when the lock becomes obsolete (let’s call this column ‘expires_at‘)

I wrote a  class that inherits from Zend_Db_Table and helps to get records with locking them.

<?php

class Koodix_Db_Table_Lockable extends Zend_Db_Table
{
    protected $_lockedByField = 'locked_by';
    protected $_expiresAtField = 'expires_at';
    protected $_TTL = '1 HOUR'; //time to live for lock

    public function fetchLocked( Zend_Db_Table_Select $select,
        $lockerID ) {

        $db = $this->getAdapter();
        $db->beginTransaction();

        $column = $db->quoteIdentifier( $this->_lockedByField );
        $select->forUpdate()
             ->where("$column=? OR $column IS NULL", $lockerID);

        $data = $this->fetchAll($select);
        if( empty($data) ) return null;

        $expiresAt = new Zend_Db_Expr('DATE_ADD( NOW(),
            INTERVAL ' . $this->_TTL . ')');
        if( sizeof($this->_primary) > 1 ) {
            foreach( $data as $item ) {
                $item->{$this->_lockedByField} = $lockerID;
                $item->{$this->_expiresAtField} = $expiresAt;

                $item->save();
            }
        }
        else {
            $arrIds = array();
            foreach( $data as $item ) {
                $arrIds[] = $item->id;
            }

            $this->update(
                array(
                    $this->_lockedByField => $lockerID,
                    $this->_expiresAtField => $expiresAt,
                ),
                $db->quoteIdentifier(current($this->_primary)) .
                    ' IN ("'.implode('","', $arrIds).'")'
            );
        }

        $db->commit();
        return $data;
    }

    public function releaseLocks( ) {

        $column = $db->quoteIdentifier( $this->_expiresAtField );

        return $this->update(
            array(
                $this->_lockedByField => null,
                $this->_expiresAtField => null,
            ),
            "$column <= NOW()"
        );
        ;
    }
}

If the table has a composite primary key (containing more than one column), the ActiveRecord approach is used, so the save() method for every record is called, that’s simple (drawback — multiple update queries). Otherwise, if it is a deep-seated table with one ID column as a primary key, then the IDs are collected in a list and all records are updated by a single statement with IN in where clause (which is much faster).

TTL (‘Time to Live‘) — period of time when lock is allowed. In my application the default is one hour. Format of TTL can be seen in MySQL documentation.

And now how to use it.

Let’s imagine you have several editors that divide the big articles list and review them. My model class has a method fetchForUser() that returns no more than 5 articles for current user (by given user ID).

This is an Article table model, inherited from the class above. Usually such classes are located at

application/default/models/ArticleTable.php

<?php
class ArticleTable extends Koodix_Db_Table_Lockable
{
    protected $_name = 'article';

    public function fetchForUser( $userId, $count=5 ) {

        $select = $this->select()
            ->where('reviewed = 0')
            ->order('expires_at DESC')
            ->order('date_imported DESC')
            ->limit( $count );

        return $this->fetchLocked($select, $userId);
    }
}

Note: if the editor refreses the page, the expres_at fields is refreshed by current time as well.

As for step four of our algorithm (releasing all obsolete locks) — create an action in your backend controller, call your table model releaseLocks() method in it and call that action periodically by Cron.

To boost the performance of the lock releasing, create an index on the expires_at column. (Because of this reason I rejected the ‘locked_since‘ column in favor of ‘expires_at‘)

P.S. In my database date/time columns have DATETIME type. If you use INT to store timestamps, convert it to unix time and back.

As certified MySQL developer (yes, I’m listed on mysql.com!!! ), I would like to share some experience I’ve got during training to the certification. Today I will tell you how to speed up your queries.

1. Use persistent connections to the database to avoid connection overhead.
2. Check all tables have PRIMARY KEYs on columns with high cardinality (many rows match the key value). Well,`gender` column has low cardinality (selectivity), unique user id column has high one and is a good candidate to become a primary key.
3. All references between different tables should usually be done with indices (which also means they must have identical data types so that joins based on the corresponding columns will be faster). Also check that fields that you often need to search in (appear frequently in WHERE, ORDER BY or GROUP BY clauses) have indices, but don’t add too many: the worst thing you can do is to add an index on every column of a table (I haven’t seen a table with more than 5 indices for a table, even 20-30 columns big). If you never refer to a column in comparisons, there’s no need to index it.
4. Using simpler permissions when you issue GRANT statements enables MySQL to reduce permission-checking overhead when clients execute statements.
5. Use less RAM per row by declaring columns only as large as they need to be to hold the values stored in them.
6. Use leftmost index prefix — in MySQL you can define index on several columns so that left part of that index can be used a separate one so that you need less indices.
7. When your index consists of many columns, why not to create a hash column which is short, reasonably unique, and indexed? Then your query will look like:

   SELECT *
   FROM table
   WHERE hash_column = MD5( CONCAT(col1, col2) )
   AND col1='aaa' AND col2='bbb';

8. Consider running ANALYZE TABLE (or myisamchk --analyze from command line) on a table after it has been loaded with data to help MySQL better optimize queries.
9. Use CHAR type when possible (instead of VARCHAR, BLOB or TEXT) — when values of a column have constant length: MD5-hash (32 symbols), ICAO or IATA airport code (4 and 3 symbols), BIC bank code (3 symbols), etc. Data in CHAR columns can be found faster rather than in variable length data types columns.
10. Don’t split a table if you just have too many columns. In accessing a row, the biggest performance hit is the disk seek needed to find the first byte of the row.
11. A column must be declared as NOT NULL if it really is — thus you speed up table traversing a bit.
12. If you usually retrieve rows in the same order like expr1, expr2, ..., make ALTER TABLE ... ORDER BY expr1, expr2, ... to optimize the table.
13. Don’t use PHP loop to fetch rows from database one by one just because you can — use IN instead, e.g.

    SELECT *
    FROM `table`
    WHERE `id` IN (1,7,13,42);

14. Use column default value, and insert only those values that differs from the default. This reduces the query parsing time.
15. Use INSERT DELAYED or INSERT LOW_PRIORITY (for MyISAM) to write to your change log table. Also, if it’s MyISAM, you can add DELAY_KEY_WRITE=1 option — this makes index updates faster because they are not flushed to disk until the table is closed
16. Think of storing users sessions data (or any other non-critical data) in MEMORY table — it’s very fast.
17. For your web application, images and other binary assets should normally be stored as files. That is, store only a reference to the file rather than the file itself in the database.
18. If you have to store big amounts of textual data, consider using BLOB column to contain compressed data (MySQL’s COMPRESS() seems to be slow, so gzipping at PHP side may help) and decompressing the contents at application server side. Anyway, it must be benchmarked.
19. If you often need to calculate COUNT or SUM based on information from a lot of rows (articles rating, poll votes, user registrations count, etc.), it makes sense to create a separate table and update the counter in real time, which is much faster. If you need to collect statistics from huge log tables, take advantage of using a summary table instead of scanning the entire log table every time.
20. Don’t use REPLACE (which is DELETE+INSERT and wastes ids): use INSERT ... ON DUPLICATE KEY UPDATE instead (i.e. it’s INSERT + UPDATE if conflict takes place). The same technique can be used when you need first make a SELECT to find out if data is already in database, and then run either INSERT or UPDATE. Why to choose yourself — rely on database side.
21. Tune MySQL caching: allocate enough memory for the buffer (e.g. SET GLOBAL query_cache_size = 1000000) and define query_cache_min_res_unit depending on average query resultset size.
22. Divide complex queries into several simpler ones — they have more chances to be cached, so will be quicker.
23. Group several similar INSERTs in one long INSERT with multiple VALUES lists to insert several rows at a time: quiry will be quicker due to fact that connection + sending + parsing a query takes 5-7 times of actual data insertion (depending on row size). If that is not possible, use START TRANSACTION and COMMIT, if your database is InnoDB, otherwise use LOCK TABLES — this benefits performance because the index buffer is flushed to disk only once, after all INSERT statements have completed; in this case unlock your tables each 1000 rows or so to allow other threads access to the table.
24. When loading a table from a text file, use LOAD DATA INFILE (or my tool for that), it’s 20-100 times faster.
25. Log slow queries on your dev/beta environment and investigate them. This way you can catch queries which execution time is high, those that don’t use indexes, and also — slow administrative statements (like OPTIMIZE TABLE and ANALYZE TABLE)
26. Tune your database server parameters: for example, increase buffers size.
27. If you have lots of DELETEs in your application, or updates of dynamic format rows (if you have VARCHAR, BLOB or TEXT column, the row has dynamic format) of your MyISAM table to a longer total length (which may split the row), schedule running OPTIMIZE TABLE query every weekend by crond. Thus you make the defragmentation, which means more speed of queries. If you don’t use replication, add LOCAL keyword to make it faster.
28. Don’t use ORDER BY RAND() to fetch several random rows. Fetch 10-20 entries (last by time added or ID) and make array_random() on PHP side. There are also other solutions.
29. Consider avoiding using of HAVING clause — it’s rather slow.
30. In most cases, a DISTINCT clause can be considered as a special case of GROUP BY; so the optimizations applicable to GROUP BY queries can be also applied to queries with a DISTINCT clause. Also, if you use DISTINCT, try to use LIMIT (MySQL stops as soon as it finds row_count unique rows) and avoid ORDER BY (it requires a temporary table in many cases).
31. When I read “Building scalable web sites”, I found that it worth sometimes to de-normalise some tables (Flickr does this), i.e. duplicate some data in several tables to avoid JOINs which are expensive. You can support data integrity with foreign keys or triggers.
32. If you want to test a specific MySQL function or expression, use BENCHMARK function to do that.

Some of these hints are unapplicable if you use a framework because direct queries are uninvited guests in the case: focus on competent database optimization — tune indexes and server parameters.

More on queries optimization:

• Optimizing SELECT and Other Statements
• Optimizing for Query Speed (more about good index creating practices).
• ORDER BY RAND()