[tech] martello speed

James Andrewartha trs80 at ucc.gu.uwa.edu.au
Wed Mar 23 18:41:43 WST 2005


The new SATA controller has been installed in martello, and I've been
benchmarking various RAID configurations with bonnie++. The results are
somewhat interesting (commands used to set up each configuration included 
in case I did something stupid):

RAID 5
------
mdadm -Cv -l5 -n4 -c128 /dev/md0 /dev/sx8/[0123]p1

(this only resynced at 22MB/s)

pvcreate /dev/md0
vgcreate reliable /dev/md0
lvcreate -L 40G -n home reliable
mke2fs -b 4096 -E stride=32 -j -m 1 -O dir_index -v /dev/mapper/reliable-home
bonnie++ -d bonnie++ -u trs80 -m martello
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
martello         4G 38175  92 65903  27 20999   8 26989  60 58108  12 170.9   0

(fiddling with blockdev --setra didn't do much)

RAID 1+0
--------
mdadm -Cv -l1 -n2 /dev/md0 /dev/sx8/[01]p1
mdadm -Cv -l1 -n2 /dev/md1 /dev/sx8/[23]p1

(these would only resync at a total rate of 40MB/s, regardless of how whether
I stopped one array or not. Further, when only md0 was resyncing, an hdparm of
/dev/sx8/3p1 returned 30MB/s, compared to the normal 55MB/s)

pvcreate /dev/md0 /dev/md1
vgcreate reliable /dev/md0 /dev/md1
lvcreate -L 40G -i2 -I32 -n home reliable
mke2fs -b 4096 -E stride=8 -j -m 1 -O dir_index -v /dev/mapper/reliable-home
bonnie++ -d bonnie++ -u trs80 -m martello
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
martello         4G 39478  97 66125  29 19881   5 29657  65 58542   8 199.8   0

Bare disk
---------
mke2fs  -j -m 1 -O dir_index -v /dev/sx8/0p1
bonnie++ -d bonnie++ -u trs80 -m martello
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
martello         4G 38044  93 57773  25 17714   6 27826  60 45065   5 215.6   0


The bare disk result bears out the hdparm value of 55MB/s. The question is,
where's the performance improvement that should happend for RAID 1+0 and
RAID 5? The slow resync and hdparm result during array construction says to
me that the ports on the controller are not independent.

A little searching finds
http://forums.storagereview.net/index.php?showtopic=17794&hl=promise which
claims "The linux driver sends only one command to the board at a time."


-- 
# TRS-80              trs80(a)ucc.gu.uwa.edu.au #/ "Otherwise Bub here will do \
# UCC Wheel Member     http://trs80.ucc.asn.au/ #|  what squirrels do best     |
[ "There's nobody getting rich writing          ]|  -- Collect and hide your   |
[  software that I know of" -- Bill Gates, 1980 ]\  nuts." -- Acid Reflux #231 /



More information about the tech mailing list