|
Hello everyone,
I recently posted to this mailing list when I
had problem with my HPT 370A add-on controller to make it work w/4 Maxtor hard
drives. Well I did not get it to work in the hpt's "hardware" raid mode (4x80GB
Maxtors in RAID 0, would work with 2 disks but not with 4), but Linux software
raid works just fine (/dev/md stuff)....I also have one other RAID 0
configuration on the same computer, which is another 2x80GB RAID 0 array
attached to motherboard's build-in controller (hdc and hdd). Ever since the
system has been rock-stable (40 days of uptime until the power outage) and also
quite performing.
I have recently done some benchmarks just to get
the ballpark idea of how fast it really is (its a server with 100mbps ethernet
and when i d/l i get 11.5MB/sec through nfs on both arrays).
I dont have the problem with the add-on card array,
which is quite fast (individual disks' hdparm -t /dev/hd[efgh] get about
35MB/sec, the array gets about 100MB/sec which i could attribute to PCI
bandwidth limitation). However I recently tested my other 2 disks' performance
and I got: (md0 == hdc & hdd in RAID 0)
/dev/hdc:
Timing buffered disk reads: 64 MB in 1.98 seconds = 32.32 MB/sec /dev/hdd:
Timing buffered disk reads: 64 MB in 1.97 seconds = 32.49 MB/sec /dev/md0
Timing buffered disk reads: 64 MB in 2.47 seconds = 25.91 MB/sec Any ideas as to why this the array is evidently
underperforming? I have already thought about the same ide channel conflicts,
the stripe size not optimized for filesystem (64k on ext3)..would any of those
qualify? If not what else could there be as the possible source of the
problem?
Thanks for any help and suggestions.
MK
|
