On occasion, I will have users complain that their computer is starting up slow. Usually the first thing that gets blamed is group policy, but I know that it usually isn’t the primary cause (just the symptom of some other issue). Often, it indicates some issue specific to that machine. How do I know that? We have spent time over the years doing some regular performance testing and timing our startups. This used to involve using a stopwatch, but it can always be difficult to get an accurate time with that.
Today, we use a tool from the Windows Performance Toolkit (and specifically XPerf) to do our testing. It will give us an accurate reading of our startup times, as well as helps us troubleshoot those systems with a slow startup. Here are some example startup times in our environment:
*WTG is Windows To Go running from a USB key
You can see from this that if a computer is at a location without a domain controller (one of our corps) the startup time increases but is still typically under 2 minutes. We do see a startup time delay increase a bit when going over DirectAccess, but it is still well under our three minute mark that we like to see.
So, how do I use XPerf to do my testing? Well, rather than retype it all, I want to share an article (that has some good additional links in it) that covers setup and testing scenarios: