Angelo's Soapbox |
|
Thursday Sep 03, 2009
Recording: "ZFS + SSD tuning for databases"
Just in case you missed the live event, we have a recording of the ZFS + SSD for databases webcast You can also download the slides from Slide share. Posted at 10:32AM Sep 03, 2009 by angelo in General | Comments[3] |
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
According to your PG benchmark putting log on SSD + data on HDD yields a higher performance than putting data on SSD.
I found that very surprising.
SSD is much faster than HDD in random IO, while in sequential IO they're similar in speed.
So it'd seem to make sense to put put the data (which require random access) on SSD. Meanwhile putting the log on SSD or HDD won't make much difference.
This MySQL benchmark bore that out: http://it.anandtech.com/IT/showdoc.aspx?i=3532&p=11
"Transactional logs are written in a sequential and synchronous manner. Since SAS disks are capable of delivering very respectable sequential data rates, it is not surprising that replacing the SAS "log disks" with SSDs does not boost performance at all. However, placing your database data files on an Intel X25-E is an excellent strategy."
Your benchmark suggested the exact opposite: put data on HDD & put log on SSD. It doesn't make sense to me even on a conceptual level. Can you explain why you got that result?
Posted by Andy on September 04, 2009 at 12:05 AM EST #
Hi Andy:
I have a theory on why this is happening but before I spill it let me do some more research and post a response blog.
-Angelo
Posted by Angelo on September 04, 2009 at 03:37 PM EST #
Hi Angelo,
Thanks. Looking forward to reading your new post.
Andy
Posted by Andy on September 04, 2009 at 04:13 PM EST #