Judging by this message in the 2nd poster's reply, it may be a case of trying to drive too much I/O or else that the arary is over taxed:
Oct 21 03:08:36 vm09 vmkernel: 54:13:51:01.362 cpu5:1029)StorageMonitor: 222: vmhba1:1:2:1 status = 8/0 0x0 0x0 0x0
The 8/0 0x0 0x0 0x0 decodes to BUSY in SCSI terminology. The first poster mentioned 1000 VMDKS also.
When you see the aborted commands coming back form the QLogic (or any) driver, it means that a command was issued from the controller to the target, but the target did not accept the i/o and therefore the i/o was aborted. When this happens, there are a few different scenarios that can occur
1) The i/o is re-queued at the driver level (HBA cache). In the QLogic and Emulex drivers, the queue size defaults to 32, but this is changeable
2) The i/o cannot be re-queued as the HBA queue is already full. In this case, a record of this is propogated back to the VM via vmkernel, and the guest (VM) will cknowledge. Then a bus reset is sent form the guest. You may see some VSCSI warnings/errors in /var/log/vmkernel at this time. You'd also expect to see eventID 11, 15,51 errors inside Windows VMs. Some Liux kernels will ignore this until a sustained period expires, at whichc stage the filesystem inside the guest will be places read-only. Windows will bluescreen if it has no disk access fter 30 seconds (see vmware KB 1014 for more details)
So, in your case, it may be that the array is over-taxed (for the second poster the 8/0 warning indicates this).
In both your cases, you might want to consider setting the HBA queue depth to 64 or even 128 (0-255 is the range) This is described in the SAN configuration guide (search for ql2xmaxqdepth, etc)
The VM's queue is fixed at a max of 128, but still - you can change the LSILogic queue depth settings inside Windows (regedit (symmpi is the key)). Linux has an adjustable queue.
Also, you might want to run esxtop again and use the j option to view disk average latency, which will tell you how long it takes for the array to respond. If this is above 100ms, you'd be looking at a sluggish array.
HTH - let me know how you get on
