A Tip For Reducing Disk Queue Length On Windows Servers

Yesterday I noticed one of the Terminal Servers (Windows Server 2008 R2) I monitor was experiencing high disk queue length (essentially the hard disk was being thrashed) by the ten-or-so users logged on.  CPU and memory utilisation were in normal ranges – in fact: there was plenty of memory to spare.

The extra memory available was due to a recent upgrade because the client was running out of memory, causing extra disk paging and slowing down the entire terminal server.  Originally I wanted to suggest an SSD upgrade for their hard disk RAID1 array, but decided with the following approach instead.

I set the page file size from system managed (which was 16GB-24GB of memory) to a set size of 4GB.  I chose this size rather than something smaller like 1GB because the users frequently run QuickBooks, large Excel spread sheets, a large Access database and large PDF files in Adobe Reader.  The users also run the usual, modern, gluttonous web browsers.

After a reboot I observed almost no disk thrashing and the hard disk queue length was reduced tenfold.  What I’d done was force the system to use more of its memory it had available.  A colleague of mine has since implemented this “trick” on another client’s terminal server with success.

This concludes my brief support article.  If this article has helped you, please let me know in the comments section below.