HP 1TB FATA Drives in EVA4400 Problems

[Dr_No]

Hello All,
Here is my issue...

So I am not 100% sure that it was the FATA drives causing the issue, but check this...

I have 8 HP BL680c G5 Server in 2x C7000 enclosure (4 servers on each). All blades are connected to the EVA4400 Dual Head Controller using QLogic FC cards). All server running VMware ESX 3.5 Update 2 All paths have been properly zoned. Everything was working ok until I installed 8x 1TB FATA Drives into my EVA. I created 4x 1TB LUN and assigned it to 4 virtual machines (DB Servers as backups). I started to noticed that I was getting "Server - No heartbeat" errors from the Vmware Virtual Center even when the servers appeared to be up. But that wasnt my only issue. I started to noticed that remote desktop and the virtual console were VERY SLOW at this point in time. I mean it would take 3-5 minutes just to open the task manager. At the beginning, I invested some time and thought the problem to be network related. So I started monitoring my network devices. I came up with nothing. So at the end, I removed the LUN from each server and used my regular 15k FC disks as backups as a test. So far it has been 3 days and no issues.

Could the FATA drives be causing this issue? Or is it a coincidence? Anyone else noticing problems with these drives?

[Uwe Zessin]

Yes, that is quite possible. Especially if the FATA-L drives do not have the latest firmware and switch to read-after-write mode when they feel that the temperature is 'too cold'.
Unfortunately, this slowness can affect other disk groups.

[Víctor Cespón]

Make sure those 1 TB FATA disks have HP06 firmware. Also, do not try to send hundreds of I/Os per second to a disk group with only 8 FATA disks. If the FATA disk group is saturated this affects also the other disk group.

Check this:

http://h20000.www2.hp.com/bizsupport/TechSupport/Document.jsp?objectID=c01671044

[Dr_No]

I upgraded the FATA drives from HP04 to HP06 today. Ill re-assigned the LUNS and see if it performs better.

Ill update the thread....

Nice link btw...
I thought vRAID-5 was suppose to have the highest IOPS?