Home/ Questions/Q 8948895
In Process

The Archive Base Latest Questions

Editorial Team
  • 0
Asked: 2026-06-15T13:06:46+00:00

I’ve hosted a WCF app in IIS 7.5 with HTTPS, TCP and Named pipes

  • 0

I’ve hosted a WCF app in IIS 7.5 with HTTPS, TCP and Named pipes bindings.

Currently working on a new version of the webservice, trying to get performance improvements, I’m testing it with client (VS console client) and server (IIS hosted) on the same server when I discover something strange:
When doing tests of 100-1000 calls to a method returning a simple string, a builtin account outperforms a custom ApplicatiolPool identity, a domain user, with about 3x the speed.

I need the AppPool identity to be an domain admin because of the things the service is required to do.

Anyone that can shed some light on this?

Share

Leave an answer

You must login to add an answer.

Need An Account,

1 Answer

  1. Editorial Team

    Seems like I have found the “problem”, even though I can’t quite understand it.

    As I mentioned I used a VS console client, i.e. debugged in visual studio. After doing some analyzing with process monitor, running both with and without builtin AppPool user, I saw visual studio doing something funky.
    When running the AppPool under a domain-admin, VS spawned an additional .vshost.exe process under a different user (couldn’t see due to seeing only SID that I didn’t bother hunting down) that shot to the roof with network load.

    I built the console-project and ran the executable outside of VS, and behold, no noticeable difference between AppPool-users and many times faster than in VS.
    No idea why this is.

    My conclusion is “When in doubt, run Process Monitor!” and that I shouldn’t have relied on the visual studio debugger so heavily for performance testing.

    • 0