How to calculate a good InnoDB log file size – recap

Following Baron Schwartz’ post: How to calculate a good InnoDB log file size, which shows how to make an estimate for the InnoDB log file size, and based on SQL: querying for status difference over time, I’ve written a query to run on MySQL 5.1, which, upon sampling 60 seconds of status, estimates the InnoDB transaction log bytes that are expected to be written in the period of 1 hour.

Recap: this information can be useful if you’re looking for a good innodb_log_file_size value, such that will not pose too much I/O (smaller values will make for more frequent flushes), not will make for a too long recovery time (larger values mean more transactions to recover upon crash).

It is assumed that the 60 seconds period represents an average system load, not some activity spike period. Edit the sleep time and factors as you will to sample longer or shorter periods.

SELECT
  innodb_os_log_written_per_minute*60
    AS estimated_innodb_os_log_written_per_hour,
  CONCAT(ROUND(innodb_os_log_written_per_minute*60/1024/1024, 1), 'MB')
    AS estimated_innodb_os_log_written_per_hour_mb
FROM
  (SELECT SUM(value) AS innodb_os_log_written_per_minute FROM (
    SELECT -VARIABLE_VALUE AS value
      FROM INFORMATION_SCHEMA.GLOBAL_STATUS
      WHERE VARIABLE_NAME = 'innodb_os_log_written'
    UNION ALL
    SELECT SLEEP(60)
      FROM DUAL
    UNION ALL
    SELECT VARIABLE_VALUE
      FROM INFORMATION_SCHEMA.GLOBAL_STATUS
      WHERE VARIABLE_NAME = 'innodb_os_log_written'
  ) s1
) s2
;

Sample output:

+------------------------------------------+---------------------------------------------+
| estimated_innodb_os_log_written_per_hour | estimated_innodb_os_log_written_per_hour_mb |
+------------------------------------------+---------------------------------------------+
|                                584171520 | 557.1MB                                     |
+------------------------------------------+---------------------------------------------+

11
Leave a Reply

avatar
11 Comment threads
0 Thread replies
0 Followers
 
Most reacted comment
Hottest comment thread
6 Comment authors
McAuleyRick JamesGeorgesonnyshlomi Recent comment authors

This site uses Akismet to reduce spam. Learn how your comment data is processed.

  Subscribe  
Notify of
Arjen Lentz
Guest

Very cool Shlomi!
Would be great to do this in 5.0 also, but that might be a shell thing since you can’t get status within a query.

sonny
Guest
sonny

GLOBAL_STATUS table is not exist in INFORMATION_SCHEMA database. using 5.1. Am I mising something…

Pls help.

George
Guest
George

Thanks for the snippet. Very handy.

What would you do if you got this result though? Seems a 36G log file might be a bad thing to have.

+——————————————+———————————————+
| estimated_innodb_os_log_written_per_hour | estimated_innodb_os_log_written_per_hour_mb |
+——————————————+———————————————+
| 38325872640 | 36550.4MB |
+——————————————+———————————————+

Rick James
Guest

That’s nice for instantaneous (one minute’s worth), but you might be in the middle of a spike or a lull. Instead, I use this (in newer versions of MySQL):

Innodb_os_log_written / (Uptime / 3600) / innodb_log_files_in_group

That gives a value for the log file size, assuming that rolling over once an hour is a good target.

I question the need for log files bigger than 4GB. I checked 213 machines here, one machine “wants” 5GB, it was the only one over 3GB.

McAuley
Guest
McAuley

Yeah, when I get a result like this:
9464024862720 9025597.4MB
it makes me question the accuracy of this kind of calc. I think looking at the “rollover” of the log files in relation to their size works better for me.

McAuley
Guest
McAuley

@Rick James, yes .. I tested your calc on several of our servers and it looks to be a lot more reasonable. Both the “uptime” and the “Innodb_os_log_written” are now part of the INNODB GLOBAL STATUS. My tests using your formula averaged about 42Mb, instead of the 902+ GIG that this other formula rendered.

Rick James
Guest

I attended Oracle’s MySQL Connect last weekend. One tidbit I picked up about 5.6 is…
There is less need now to avoid extra-large log files. Apparently “recover” has been made more efficient. Also, log files can now be bigger than 4GB.

More details:
http://dev.mysql.com/doc/refman/5.6/en/innodb-parameters.html#sysvar_innodb_log_file_size

No need to every have more than 2 log_files_in_group.

trackback

[…] was configured used a large value and it was annoying me  a little. After fiding the Shlomi Noach blog entry with a good query to check the size of transactions that populate the log buffer, its seems very […]