Huge drop in performance of RAID 0 measured by ATTO

[6BQ5]

I got a hold of three 15K 600 GB SAS SFF hard drives, part number 759548, to install in the 8-bay SFF cage already in my G8 server. In past posts I was ready to swap out the SFF system for the LFF design but had a change of heart once I got mu hands on these three SFF drive. They are really small and super cute!

I read that RAID 0 offers the highest performance but with no redundancy and with no fault tolerance. That's ok. I just want to see wickedly fast R/W speeds so I created a 3 x 600 GB RAID 0 array. The resulting logical disk is 1800 GB. My chosen benchmark program to measure performance is ATTO. The benchmarked results it gave me are disappointing.

ATTO wrote and read a 256 MB file using various block transfer sizes between 512 bytes and 64 MB. As you can see performance steadilt climbed up to a block size of 512 kB before taking a massive (huuuge!) tumble with 1 MB block size. Performance was pretty flat after that.

Why the drop? What could be causing this?

I also tried a 2 x 600 GB RAID 0 array and the result was the same ,,, like, identical! Shouldn't a 2 x 600 GB be slower than a 3 x 600 GB?

Is just the normal performance of this drive or the on-board controller? Am I missing some setting somewhere?

Thanks!

[SanjeevGoyal]

Hello,

Server supposed to be updated with the latest bios and all other firmware to achieve better performance. Performance does not depend only for raid configuration.

You have to especially verify what is the current SA controller and hard disk firmware.

If you feel this was helpful please click the KUDOS! thumb below!   

Regards,

[6BQ5]

Hi Sanjeev,

My BIOS version is P72, which I think is the latest for the ML350p Gen8.

My RAID controller is the Smart Array P420i. The hardware revsion is B, firmware version is 8.00. Now I see the controller memory module size is ... 512 MB.

The size of that memory module is probably what is killing me, isn't it?

[SanjeevGoyal]

Hello,

As suggested in earlier community cases, I would suggest contact to the HPE case log team for deep drive analysis.

They will collect required logs ( AHS and ADU report) for further investigation and will suggest a further action plan.

If you feel this was helpful please click the KUDOS! thumb below!   

Regards,