Hi,
This is an SBS 2008 SP2 running as a virtual machine with 12GB memory allocated and a page file of 18GB. The Hypervisor is an HP Proliant 350 G5 connected over iSCSI to a Qnaps chassis with four drives configured as RAID 10. The SBS VM is located on the Qnaps Chassis.
Periodically the server becomes unresponsive with 100% of the memory consumed and can only be shut down from the Hyper-V manager. This is starting to happen more frequently even with the server lightly loaded.
There appear to be two problem areas. One, an excessive number of page faults and two, some excessive disk activity. I have been looking at the memory issues first as I suspect that at least some of the disk activity is due to the page faults.
Microsoft Performance Monitor shows high spikes of page faults and I understand that this to be the sum of hard and soft page faults. I have used the instructions to try and differentiate between the two but the results suggest they are soft page faults. I have my doubts, otherwise why the high disk activity?
The monitoring was done over the weekend with only a light load on the server. My understanding of a soft page fault is that the data is most likely to be in memory, just not where expected. I could understand this if there was a lot going on but there isn’t. Given that this is a new area for me I suspect I am reading the results incorrectly and I really have a predominance of hard page faults.
I am also finding that the data transfer rates can be as little as 5MB/s which is lower than I would expect by a factor of 10. This may or may not be relevant but I have included it in case it is.
Can anyone give me some guidance on what are the best counters to use in Performance Monitor to identify the cause of the problem?
Regards Tegglet
