LSI SAS2008 Performance - GA-7PESH2 - RAID10

Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.

Mark Linton

New Member
Sep 28, 2018
15
1
3
So I've built a new server with the GA-7PESH2 as the basis for the system thinking that the onboard LSI SAS2008 controller would be good enough for my needs and I'm a little underwealmed by the current performance.

Hardware - GA-7PESH2 MB, LSI SAS2008 controller, one of the breakout cables is connected to 4 seagate 8TB IronWolf ST8000VN0022 drives
The drives are configured in a RAID10, thinking this would maximize performance and minimize disk failure issues.
Running vSphere 6.7 - latest patches installed to a separate SSD
Debian Linux OS running in a VM with a 7TB disk

Ran HDparm and DD to try to get some disk write stats on the drive:

root@CorePlex:/tmp# dd if=/dev/zero of=/mnt/media/test1.img bs=1G count=1 oflag=dsync
1+0 records in
1+0 records out
1073741824 bytes (1.1 GB, 1.0 GiB) copied, 7.27152 s, 148 MB/s
root@CorePlex:/tmp# dd if=/dev/zero of=/mnt/media/test1.img bs=1G count=1 oflag=dsync
1+0 records in
1+0 records out
1073741824 bytes (1.1 GB, 1.0 GiB) copied, 7.6482 s, 140 MB/s
root@CorePlex:/tmp# dd if=/dev/zero of=/mnt/media/test1.img bs=1G count=1 oflag=dsync
1+0 records in
1+0 records out
1073741824 bytes (1.1 GB, 1.0 GiB) copied, 7.35856 s, 146 MB/s
root@CorePlex:/tmp#
So 140 MB/s is not bad and seems about average for a single drive, and because the usecase for this drive is read-speed focused (media storage) I was hoping the read speeds were a little better than the following:

root@CorePlex:/tmp# hdparm -tT /dev/sdb​

/dev/sdb:
Timing cached reads: 9554 MB in 1.99 seconds = 4793.94 MB/sec
Timing buffered disk reads: 374 MB in 3.04 seconds = 122.88 MB/sec
root@CorePlex:/tmp# hdparm -tT /dev/sdb

/dev/sdb:
Timing cached reads: 8738 MB in 1.99 seconds = 4383.27 MB/sec
Timing buffered disk reads: 724 MB in 3.97 seconds = 182.39 MB/sec
root@CorePlex:/tmp# hdparm -tT /dev/sdb

/dev/sdb:
Timing cached reads: 8328 MB in 1.99 seconds = 4177.65 MB/sec
Timing buffered disk reads: 358 MB in 3.04 seconds = 117.73 MB/sec
This seems mediocre at best - I would expect that the drives would increase the speed when reading files across each of the striped pairs no? What might the issue be? The controller? the drives? VM?

Thanks.