Storage Boards Cleanup
To make it easier to find information about HPE Storage products and solutions, we are doing spring cleaning. This includes consolidation of some older boards, and a simpler structure that more accurately reflects how people use HPE Storage.
Disk Arrays
cancel
Showing results for 
Search instead for 
Did you mean: 

SmartArray 5302 Performance

Bill Gillespie
Occasional Contributor

SmartArray 5302 Performance

We have 2 different servers, an ML 370 and a ML 560 (I think). One has one 5302-64 and the other has 2 5302-128 controllers. The server with one controller has 12 internal disks, the other has 2 shelves of 14 disks per shelf with a channel connected to 7 disks (dual bus shelf). All the disks are 10k 72GB. The internals are Ultra3, and the others are Ultra320. We are running Linux.

It appears that we can get only about 70 MB/sec per channel on either box. We have tried RAID 0, RAID 0+1, and RAID 5. With just a one disk RAID 0 we can get 50 MB/sec reads. However, it appears that adding more disks to the RAID set tops out at 70 MB/sec no matter how many disks are used.

We can not tell if this is a controller bottleneck (70 MB/sec per channel) or a Ultra 160 SCSI bottleneck on the SCSI bus itself.

Has anyone gotten better performance from this card?

Also, I am wondering if the new 6402 controller that is Ultra 320 will really do any better. I think the disk shelf we are using (4354R) is an Ultra 160 device, so I am not sure the faster controller will actually produce any faster throughput.
1 REPLY
Tim D Fulford
Honored Contributor

Re: SmartArray 5302 Performance

Bill

You did well to get 50MB/s from One disk!! I think that is the max listed speed.

Anyway.... I can't say for sure if the SCSI is maxed out BUT I think it is the SCSI card.

the problem is that there is obviuosly more than one place for the bottleneck to occour - it could also be the internal bus of the storage unit. To show where it is would require some testing (sorry) during the test look at the utilisations of the disks, also daisy chain the storage units together, from these 3 tests you will be able to deduce where the bottleneck really is...you might get something like the following

1 storage unit (1x SCSI bus)
1x dsk RAID1 - 50MB/s disk util 100%
2x dsk RAID1 - 70MB/s disk util 70%
4x dsk RAID1 - 70MB/s disk util 35%
etc

2 storage unit in parallel (2x SCSI bus) disks in alternating units
1x dsk RAID1 - 50MB/s disk util 100%
2x dsk RAID1 - 100MB/s disk util 100%
4x dsk RAID1 - 140MB/s disk util 70%

2 storage units in serial (1x SCSI bus) disks in alternating units
1x dsk RAID1 - 50MB/s disk util 100%
2x dsk RAID1 - 70MB/s disk util 70%
4x dsk RAID1 - 70MB/s disk util 35%

anyway, If you are up for it those are the tests you could do.

regards

Tim

-