Analyze boot performance - With FastBoot enabled vs Without


I'm looking for comparing the boot time for machines with FastBoot enabled vs themselves (or others) when FB was disabled. For that, I'm looking at "On/Off Transition - Standby/Resume Duration - Resume from Standby(s) - Target State  = shutdown". My idea is to build a query that will compare these events with events from "On/Off Transition - Standby/Boot Duration - Computer Startup(s)".

Any other suggestion? If I'm on the right track, any hint on how I can cross these 2 sourcetypes to build my query?


  • Avatar
    Martin Kretzschmar Official comment

    Hi Gustavo,

    The problem with Windows' fast startup feature in Windows 10 is, that instead of a real reboot, it just puts Kernel and Session 0 to sleep. The uberAgent log makes this easy to understand - the service is not restarted in such a case, but continues to run.

    For the shutdown analysis, uberAgent uses the trace file referenced in the following registry path: HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Diagnostics\Performance\ShutdownCKCLSettings. In case of a fast startup, Windows will not generate those files and as a consequence, the On/Off Transition dashboards will contain no information.

  • 0
    Ary Gustavo Zimmermann

    Hi Martin,

    Question 1 - In order to capture the time our machines take to go from the turn on up to the moment where the login screen shows up, with FS enabled, we are looking at "OnOffTransition:StandbyDetail2 / ResumeFromStandbyMs". You are saying this metric is not what we are looking for?

    Question 2 - With FS disabled. We are trying to capture the time our Windows 10 machines take to go from the turn on up to the moment where the login screen shows up. For that, we are looking at "OnOffTransition:BootDetail2 / ComputerStartupMs". By doing some "stop watch" tests, I noticed that this metric counts up to the point where the screen "Applying Computer Settings" shows up, but there's a long way from the time this screen appears up to the point where the login screen shows up. Please find some observation tests below:


    FS Enabled

    Source Type / Field

    Boot Time (UA)

    Boot Time Observed

    6/16 - 9:15AM CST


    BootDetail / ComputerStartupMs

    9832 (9.9s)

    36000 (36s)

    6/16 - 10:35 AM CST


    BootDetail / ComputerStartupMs

    9957 (10s)

    36800 (36.8s)

    6/16 - 10:50 AM CST


    BootDetail / ComputerStartupMs

    9642 (9.6s)

    35390 (35.9s)


    Boot Time UA reflects the ComputerStartupMs metrics.

    Boot Obserbed Time reflects the stop watch time observed.

    Would you have any advice as to how we can capture the "entire" boot time without FS enabled, from the power on action up to the login screen appearence?



  • 0
    Martin Kretzschmar

    Hi Gustavo,

    Question 1 - The sourcetype you are using is correct for Standby/Resume transitions. I just wanted to point out, that fast startups will not show up in our Boot Duration dashboard. Sorry for the confusion.

    Question 2 - We are not able to measure the duration starting from the actual power button press. The earliest possible event, is with the beginning of the Main Boot Path phase, according to Microsofts Windows On/Off Transitions Solutions Guide. The ComputerStartupMs metric (documentation) represents the point in time when the LogonUI.exe process is launched. To validate the difference you could compare this timestamp with the one from the Windows Eventlog when the processing of computer policies finished, which can be found under Microsoft-Windows-GroupPolicy/Operational.

    Kind regards, Martin

Please sign in to leave a comment.