Q: Should I change my Hyper-V Windows Server 2012 box from system managed pagefile to a manual configuration?

A: Early in the Windows Server history, the algorithms used by the system to determine the pagefile size were purely based on the amount of memory, such as 1x or 1.5x the amount of RAM installed. Therefore, a Hyper-V server with 180GB of memory would have a pagefile consuming 180GB of disk space, which is largely unnecessary because the actual host operating system isn't using most of the memory—the memory is being used by virtual machines. Thus it was common to modify the pagefile configuration to a static value such as 8GB so as not to waste huge amounts of disk space but to still ensure that at least a kernel memory dump could be performed in the event of a crash.

Windows Server 2012 changed this; the system managed pagefile is intelligent and by default will ensure that the pagefile is at least big enough to capture a kernel memory dump but isn't the size of the complete memory. If the system experiences a bug check (blue screen), then for the next four weeks the pagefile will be increased in size to match the amount of memory. This ensures that if a bug check occurs again, the entire contents of memory can be captured; after the four weeks has passed (the date of the last bug check is stored in the registry value HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\CrashControl\LastCrashTime), the pagefile will shrink back to the smaller size. Jeff Hughes documents this in the blog post "Windows 8 and Windows Server 2012: Automatic Memory Dump."

Because of this intelligence, the best practice is now to leave the default of system managed pagefile unless you really need to ensure a smaller pagefile and are willing to accept the reduction in troubleshooting information in the event of a problem, as well as possible system instability if more memory is needed and the pagefile isn't allowed to grow.