Thanks for all the replies, Further info:
.Re:transactions That shouldn't matter. The baseline numbers show the real processing incurred in executing the trx batch when accessed directly. That is without any webservice or remoting overhead involved. .Re:Client Same platform. As I said before, everything is contained in a single machine. Client application is a console application (although I thought that wouldn't matter... does it?). .Re:anonymous access Yes. I know that the numbers did not come out very well formatted :) but they compare the *same* trx batch with anonymous and IWA. And they (painfully) show that IWA is the culprit here. I have repeated them below with a (hopefully) better fomatting. .Re:users I'm using domain accounts. Client does for web services: proxy.Credentials = System.Net.CredentialCache.DefaultCredentials and for remoting: IDictionary Props = ChannelServices.GetChannelSinkProperties(obj); Props["credentials"] = CredentialCache.DefaultCredentials; BTW, PreAuthenticate has no visible effect whatsoever... Execution results in secs: BaseLine TRX1=2,433 TRX2=18,797 WebService with Anonymous TRX1=9,273 TRX2=29,142 WebService with Integrated TRX1=80,246 TRX2=108,386 RemHTTP/BIN with Anonymous TRX1=3,315 TRX2=30,444 RemHTTP/BIN with Integrated TRX1=75,328 TRX2=106,713 RemHTTP/XML with Anonymous TRX1=4,647 TRX2=50,683 RemHTTP/XML with Integrated TRX1=81,427 TRX2=123,067 RemTCP/BIN with Anonymous TRX1=3,094 TRX2=33,278 Anonymous numbers are a pain but IWA is... oh well, you can read above. Rgds, Manuel You can read messages from the DOTNET archive, unsubscribe from DOTNET, or subscribe to other DevelopMentor lists at http://discuss.develop.com.
