Poor Performance with SAS drives???

Woodster
Woodster used Ask the Experts™
on
Currently I am testing performance of a DSN-2100-10 DLink SAN.  I have not used IOMeter before this however testing of this SAN has brought to light some interesting perfomance issues.

I have run IOMeter on our server drives and found the following:

Altos R720 Server
2 x Xeon 5130
Windows Server 20087 64bit
8GB RAM

Physical Drive 0
RAID 1 (hardware on mother board)
2 x 72GB SAS drives

Physical Drive 1
RAID 5 (hardware on main board)
3 x 1TB SATA drives

Before anyone replies that I should be using a RAID addon card etc..  The performance of the C: is a LOT slower than I would expect (which in turn is causing poor results in my testing of the SAN performance)

IOMeter Test settings
Maximum Disk Size: 2500 Sectors
Transfer Request Size: 1MB
Percent of Access Specification: 100%
Percent Read/Write Distribution: 100% Read
Percent Random/Sequential Distribution: 100% Sequential

C: Test
67GB (100%) Primary Partition on physical drive 0
Test results: approx 100MB/sec

D: Test
500GB Primary Partition on physical drive 1
Test results: approx 270MB/sec

I have never used IOmeter before and never had to do any disk performance testing but it is concerning that expensive SAS drives in a RAID 1 configuration are being out performed by cheaper SATA drives in a RAID 5 configuration.

I am chasing any pointers / information on how to get the SAS drives returning the expected performance in the above configuration.
Comment
Watch Question

Do more with

Expert Office
EXPERT OFFICE® is a registered trademark of EXPERTS EXCHANGE®
Are you using the correct connection.

there are different cables.

http://www.cs-electronics.com/sas-cables.htm

CT

Author

Commented:
Without pulling the server apart - The SAS and SATA drives are all connected to the backplane in the server via the server hot swap bays.  This would have come as configured by Acer when the server was initially purchased so I would assume that the cables are correct.
Top Expert 2014

Commented:
You're doing 100% sequential reads; high capacity SATAs are very good at this because of the high areal density means the bits come off a single track very fast. The same high areal density coupled with less powerful actuators makes them poor at random access because the head has to settle time. Try 50:50 read/write with small transfer size and make sure you make the test file very big to simulate long seeks (or format into 4 slices and use one test file on each simultaneously 2500 sectors isn't going to test much more than the disk cache.

Also you're testing on the C: drive where your OS is on, that completely invalidates the test, it's probably fragmented so the sequential test isn't sequential at all and it may also be doing other work such as logging the benchmark test results.
Acronis in Gartner 2019 MQ for datacenter backup

It is an honor to be featured in Gartner 2019 Magic Quadrant for Datacenter Backup and Recovery Solutions. Gartner’s MQ sets a high standard and earning a place on their grid is a great affirmation that Acronis is delivering on our mission to protect all data, apps, and systems.

Author

Commented:
Had planned to get back to this sooner than this and have this question closed by now.  I have been unable to find anything easily downloadable that will allow me to boot from something like a flash drive and then run an IOMeter test to get a valid test of my system drives.  Obviously I SHOULD be getting faster speeds than I am on SAS drives.  Is ther any easy way of performing a VALID performance test on my server system drive?

Author

Commented:
I have run some other tests by simply copying files from the SATA volume to the SAS volume but am still getting a lot slower than expected speeds to the RAID1 SAS volume. Copying large files sits at anywhere from 60-70MB /sec so I am still unsure how to go about getting expected access speeds to this SAS volume.
Can you check to see if wrie cachae is check on the drive by right click on the drive and under properties menu on the drive.

CT

Author

Commented:
Yes.  The Volumes are set for "Optimize for performance" so write caching IS enabled.  Both the SATA RAID5 volume and the SATA RAID1 volume have this set.

Author

Commented:
I am still not getting any better read/write speeds to these drives.  Any other pointers to possibly resolving the speed issue and getting the (supposedly high speed) SAS drives anywhere near the speed of the SATA drives in the same system

Author

Commented:
The suggested comment has not been accepted as it does not provide advice as to why the SAS drives would be so slow.  The speed test has been run with the OS running on the RAID5 SATA drives rather than the RAID0 SAS drives and has provided the same result as before.  Surely if (as suggested) that invalidated the test the SATA drives would have showed some sort of slowdown which they have not.  That comment has been disregarded for that reason.

The other comments about checking write caching and correct cables assume that something other than a basic default setting or confiuration has been changed. ie: the maufacturer (Acer) providing incorrect cabling within the server or that we had at some point turned off write caching (default setting is ON).
Top Expert 2014

Commented:
The only way you're going to get a valid test on your system disks is for them not to be your sustem disks, i.e. boot off a USB stick.
I think it would be a drivers or firmware issues.

CT

Author

Commented:
See my previous comment.  The system was moved to the RAID5 SATA disks rather than the RAID0 SAS disks and the result was the same so I have already done that.
Top Expert 2014

Commented:
Well if you wiped the SAS disks to remove any fragmentation then it's down to you doing 100% reads which isn't a very meaningful test unless it's going to be video server, or drivers as CT suggests.

Author

Commented:
We have tried 50% read and 50% write tests and have the RAID1 SAS drives still coming in at less than half the speed of the RAID5 SATA drives.  We have also tried 50/50% sequential/random and 100% random.

We have the latest drivers and firmware according to the Acer site for the Acer Altos R720 server.  Would drivers/firmware affect only SAS drives considering that both SAS andSATA drives are all connected to the same physical backplane within the server.
It sounds like the Server controller does not know how to operate the SAS drives correclty.
I wouild be it you put in a Raid contrlller  the drives would wake up and run.

CT

Author

Commented:
Unfortunately without any other updated firmware and/or drivers there is no way I can say it is or isn't the drivers and no other ideas have been thrown up so I'll close this one off.  Currently don't have an additional RAID controller at my disposal to test that further.

Do more with

Expert Office
Submit tech questions to Ask the Experts™ at any time to receive solutions, advice, and new ideas from leading industry professionals.

Start 7-Day Free Trial