Hi, I know a hard page fault in Windows is when a page referenced is no longer in physical memory. I have a w2k3 file/printer server with an attached SAN (25TB) which is showing a huge page/fault rate, an average of 6000 page fault/sec according to Windows Perfmon. According to third party tool, there are 3500 hard faults/sec, which is really worrying to me. Users aren't complaining at this moment about performance, but I think this is quite disturbing. I had a look at Process Explorer from Sysinternals and can't see any process showing up high values in the Page Faults Delta Column. How could I find out the cuplrit fo that hard fault page rates? Thank you.

you can give the xperf tool a try (http://blogs.technet.com/b/askperf/archive/2008/06/27/an-intro-to-xperf.aspx). Hard Faults tracing would give you an idea about paging.

If the Page fault/sec is high, it means there's no enough memory to support the working sets of the processes running on the server. Use the performance tool to check te traces see if there's any clue. And this article is for your reference: Investing Disk Paging http://technet.microsoft.com/en-us/library/cc938588.aspx Regards, MiyaMiya Yao TechNet Community Support

If the Page fault/sec is high, it means there's no enough memory to support the working sets of the processes running on the server. Use the performance tool to check te traces see if there's any clue. And this article is for your reference: Investing Disk Paging http://technet.microsoft.com/en-us/library/cc938588.aspx Regards, MiyaMiya Yao TechNet Community Support

you can give the xperf tool a try (http://blogs.technet.com/b/askperf/archive/2008/06/27/an-intro-to-xperf.aspx). Hard Faults tracing would give you an idea about paging. Thank you for the link. I knew about that tool but it's being a long time since my last experience with it. I ran the command xperf -on base to have a look at the I/Os because the read/write rate is quite high (800 reads/sec for instance) taking into account the average number of users connected to the server. The trace shows up $mft with the most i/o time by far. Does it have anything to do with a defragmented disk? Thank you.

Hmm, as you speak of a 25TB attached SAN and say its a fileserver, how big is your $MFT (analyse defrag run should report it)? If its very large, it could deplete Paged Pool Memory as I understand this article (http://blogs.technet.com/b/askperf/archive/2011/09/23/getting-to-know-the-mmst-pool-tag.aspx - Determining Root Cause part. Another article on values of pool memory http://blogs.technet.com/b/askperf/archive/2007/03/07/memory-management-understanding-pool-resources.aspx) which would lead to paging.

Having a look at the root causes: - There are a few pst files that are being deleted from the Volume. In fact, there are a lot of them, but most of them are mail backup files from the users. - Heavy File I/O. Server hangs when I/O raises, but the RAID is made of 14 SAS drives and the server collapses when iops goes to 500. I don't think that a 14 SAS disks RAID 5 can't coup with 500 iops or more. - Large $MFT. This is my favourite culprit. I'll taka advantage of PoolTag from OSR. Thank you.

Having a look at the root causes: - There are a few pst files that are being deleted from the Volume. In fact, there are a lot of them, but most of them are mail backup files from the users. - Heavy File I/O. Server hangs when I/O raises, but the RAID is made of 14 SAS drives and the server collapses when iops goes to 500. I don't think that a 14 SAS disks RAID 5 can't coup with 500 iops or more. - Large $MFT. This is my favourite culprit. I'll taka advantage of PoolTag from OSR. Thank you.