Inconsistent bonnie++ Output Across Various RAID Configurations
I have 3 RAID configurations on a Debian 3.0 system running 2.6.10 #6
SMP:
a 2-disk ext3 RAID 1, for system files
a 4-disk JFS RAID 10, intended for PostgreSQL WAL
a 12-disk JFS RAID 10, intended for postgres data
The two RAID 10s are managed by MegaRAID cards.
Once the box was configured, I ran through a few iterations of bonnie+
+ just to test raw I/O of the new disk subsystem. I got some
confusing results.
Here are the results from the 4-disk RAID 10:
bonnie++ -b -s 16g
Version 1.03 ------Sequential Output------ --Sequential Input-
--Random-
-Per Chr- --Block-- -Rewrite- -Per Chr- --
Block-- --Seeks--
Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %
CP /sec %CP
foo 16G 47196 97 60658 21 41152 8 44940 85 120093 11
487.5 0
------Sequential Create------ --------Random
Create--------
-Create-- --Read--- -Delete-- -Create-- --
Read--- -Delete--
files /sec %CP /sec %CP /sec %CP /sec %CP /sec %
CP /sec %CP
16 31662 84 +++++ +++ 28794 99 14897 96 +++++ ++
+ 18102 89
For one thing, what are the conditions under which bonnie++ will fail
to provide output for the file creation tests?
Here are the results from the 12-disk RAID 10 (with the same command
line options):
Version 1.03 ------Sequential Output------ --Sequential Input-
--Random-
-Per Chr- --Block-- -Rewrite- -Per Chr- --
Block-- --Seeks--
Machine Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %
CP /sec %CP
foo 16G 20164 41 24381 7 82487 16 46214 87 161688 13
329.5 0
------Sequential Create------ --------Random
Create--------
-Create-- --Read--- -Delete-- -Create-- --
Read--- -Delete--
files /sec %CP /sec %CP /sec %CP /sec %CP /sec %
CP /sec %CP
16 1436 3 +++++ +++ 653 1 624 3 +++++ ++
+ 702 2
The results for the 2-disk RAID were slightly faster than but more
similar to the 12-disk RAID than to the 4-disk RAID.
I can't for the life of me figure out why the creation and deletion
times for the 12-disk RAID are so abysmal compared to the 4-disk
RAID. I tried the tests repeatedly, also throwing in -n
128:20000:16:512 thinking that 0-byte files might have something to
do with it, but the file creation performance profiles stayed roughly
the same.
Is this performance pattern cause for concern? File creation isn't
the primary function of a database server, but I'm more worried that
this is indicative of a problem with the RAID configuration or
something else at the filesystem level.
--
Thomas F. O'Connell
Database Architecture and Programming
Co-Founder
Sitening, LLC
http://www.sitening.com/
3004 B Poston Avenue
Nashville, TN 37203-1314
615-260-0005 (cell)
615-469-5150 (office)
615-469-5151 (fax)
Reply to: