I am having problems with a website with 600MB of MySQL database. The website is way too slow. I noticed the bigger the MySQL database gets, the slower it goes. When it was 5MB, the website was very fast. When it started getting bigger it started to get slower and slower and now, at 600MB, it is really slow, taking like 10 seconds to load pages.

I checked the top processes and it is nothing related to high load or anything. It is not even related to IOPS as I tested on HDD 7.2k rpm drives and it gave same problem now with testing with Intel 320 SSD drives, so I don't think it is about high queries either.

The website is using Wordpress and there are like 9 plugins active. People said it might be the plugins...well maybe...but right now I just want to cache the whole database in memory and would like to get help and direction on where to start and how to do it.

I have 16GB RAM and i5-2400 4 cores @ 3.1 GHz. OS is centos 5.7

top - 07:23:57 up 9 days, 12:15, 0 users, load average: 0.09, 0.04, 0.05
Tasks: 162 total, 1 running, 161 sleeping, 0 stopped, 0 zombie
Cpu(s): 8.2%us, 1.0%sy, 0.0%ni, 90.7%id, 0.0%wa, 0.0%hi, 0.0%si, 0.0%st
Mem: 16367532k total, 3641628k used, 12725904k free, 612140k buffers
Swap: 1046520k total, 0k used, 1046520k free, 1538896k cached

3 Answers 3


If I were you, I would switch all data to InnoDB. Table locking/row locking has long been discussed by many. I would always choose InnoDB hands down. However, there is another profound reason for choosing InnoDB...CACHING.

While most people boast that MyISAM is faster for reads, most people forget that the many cache for MyISAM, which is called the key cache (set by key_buffer_size), only caches index pages from .MYI files. It never caches data pages. It has an official maximum of 4GB in 32-bit Systems. 8GB is best maximum for 64-bit.

The InnoDB Buffer Pool caches the data and index pages. Depending on the server your have, you can cache up to the entire dataset in RAM. You can tune InnoDB for up to 80% RAM and 10% for DB Conenctions, and leave 10% for the OS. This is true even for different operating systems.

I have recommended these things for Drupal customers with marvelous success. It applies to Wordpress just as well. I have provided DB support for clients with WordPress. Same improvements.

You can always configure memory for InnoDB more effectively that you can more MyISAM. There is always a way to tweek InnoDB to suit your performance needs. As your data grows, it will eventually become a requirement.

UPDATE 2011-11-21 11:44 EST

If your complete dataset is small enough, you could execute a SELECT query on every table you have right after mysql starts up.

For all tables that are InnoDB and/or MyISAM, run this query:

    CONCAT('SELECT ',ndxcollist,' FROM ',
    db,'.',tb,' ORDER BY ',ndxcollist,';') SelectQueryToLoadCache
        engine,table_schema db,table_name tb,index_name,
        GROUP_CONCAT(column_name ORDER BY seq_in_index) ndxcollist
    FROM (
            information_schema.statistics A INNER JOIN
            (SELECT engine,table_schema,table_name
            FROM information_schema.tables
            WHERE engine IN ('InnoDB','MyISAM')) B
            USING (table_schema,table_name)
            B.table_schema NOT IN ('information_schema','mysql')
            AND A.index_type <> 'FULLTEXT'
        ORDER BY
        ) A
) AA
    engine DESC,db,tb

This will output every possible SELECT query you need to run that will summon all indexes to be referenced. Place this query in a file called /root/MakeSelectQueriesToLoad.sql. Run the script and collect the output /root/SelectQueriesToLoad.sql. Finally, run it:

mysql -u... -p... -AN < /root/MakeSelectQueriesToLoad.sql > /root/SelectQueriesToLoad.sql
mysql -u... -p... < /root/SelectQueriesToLoad.sql

This will definitely preload all index pages into the InnoDB Buffer Pool and MyISAM Key Cache. If all your data is InnoDB, make two changes:

  • replace WHERE engine IN ('InnoDB','MyISAM') with WHERE engine='InnoDB'
  • replace CONCAT('SELECT ',ndxcollist,' FROM ', with CONCAT('SELECT * FROM ',

This will also populate more data pages into the InnoDB Buffer Pool.

FINAL NOTE : Make sure the InnoDB Buffer Pool is large enough to hold all your InnoDB Data

  • This can fail if the idio^Wperson that created the database used reserved words as any of the column names. You can edit this manually, or you can have it detect reserved words and quote them in the result. I use a WITH table to hold all the reserved words, but you can also create a table to hold them for the same result. There was mysql.help_keyword, but it hasn't been supported since 10.2.7, apparently, and doesn't exist at all in 10.5.5. Commented Jan 25, 2021 at 3:55

You already are caching the whole database in memory. The problem is almost certainly the time it takes to search the database, even in RAM.

Watch your disk I/O stats. You'll probably see that there's just the occasional random bit of disk I/O. The database is in memory. That's not the problem. You need to install iostat first. You don't mention your platform or distribution, but it's probably in a package called iostat. You may find atop more friendly.

Did the people who told you that do so after getting any evidence that your entire database wasn't already in memory or that disk I/O was the problem? Otherwise, their advice is the equivalent of a doctor who has never seen or examined you but just heard that your arm hurt telling you to put a cast on it.


Install a good caching plugin for Wordpress, it might help. But sooner or later you have to figure out the bottleneck that slows your system.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.