IOPs on storage device is pegged by VMware View Desktop

[The Picture]

It has a strange trouble with just one desktop pools that a lot of services. This particular desktop group has about 250 desktop session going. From time to time all the way through the day, one or two of the desktops in the pool that the parity of the underlying data store / device, to the point of reading and writing tens of thousands of stroke IOP hundred (10,000/250,000 IOPS of writing). This makes the rest of the data storage systems in particular to reduce drastically. The problem is alleviated when a particular system user logs off and the desktop is renewed (This is a non-Persistent Desktop pool).

Liquidware Labs shows nothing unusual, that is, the guest operating system is not likely that the potential cause of the trouble. I am unable to detect the trouble during flight due to the fast way users log in to check ESXTop live is not possible at this time.

What seems to be the appearance is the core of the guests is causing extreme IOP without the guest actually able to observe (Liquidwarelabs). I'm thinking of a driver or something in the VMX file might be causing this

[Diamond!]

I have never seen vmx difficulty. The major storage troubles I have ever seen have resulted from AV dat downloads or exploration. Is it the identical desktop or desktop computers in a different way each time?

[Alfons]

I have never seen vmx difficulty. The major storage troubles I have ever seen have resulted from AV dat downloads or exploration. Is it the identical desktop or desktop computers in a different way each time?

AV scans and updates are off. AV is only set up for asset protection. It seems quite clear that a kernel problem that I cannot understand.

[Ooghavanth]

What hard drive controller is used (pvscsi?) And Network Controller (vmxnet3?) Just to see if there is something there.

He has also gone through optimization guide for Windows 7? There are a few tasks that run Windows at odd times, also some components that can run on the hard drive as if you’re creating a restore point. But 10000-250000 is a lot of IOPS, SSDs are running?

[Parnell]

Might be using the desktop all the RAM accessible so that paging occurs.This would cause a set of disk beating; also make use of something like Altirs patch.

[Erakna]

We are using LSI logic controller parallel and flexible network adapter (not VMXNET). This fund contains the Windows XP desktop virtual machines.

There is a patch on the delta of the RV, only the teacher, and for that we make use of invited VUM patch still appears in the control of vSphere 4.1.Storage IO would be an option for virtual machines do not break the group, nevertheless not a solution to the main trouble.

[sHaRiCe]

How long the problem is caused by? If it is more than 5 minutes, should appear on Stratusphere (Liquidware Labs). Does the machine user account that (apart from the general slowdown of the machine)? I log a ticket and ask for their help LiquidwareLabs to inspect and examine the reports of the periods in which this has occurred to perceive if something can find help.

Nothing in theVCenter is really only going to solve this problem, you need to know what process is causing the storm E / S. AV have been said - do not use AV updates? Not so bad? AV Provider What is it? Symantec has a particular set of configuration requirements and improved version that are useful for VDI environments. Do users have access to Internet from your VDI session? Could be malware of some sort?

[JyotiT]

Anyway, give LiquidwareLabs a call and see if they can help reduce a process if windows or third parties that is causing this. Also look at firewall logs of the affected virtual machine to determine which websites were in, look for trends between 2 or 3 events. You may get lucky. Is it the same user each time .even more interesting!