The interesting (and by “interesting” I mean completely frustrating) thing about networking is that it involves so many pieces of hardware and software. Adapters, routers, firewall devices, operating system components, third-party tools, applications…the list goes on and on.
If you’re going to make sense of this sort of system, it helps to start with a reliable baseline, which is where the Internet Connectivity Evaluation Tool comes in. Visit this page using Internet Explorer on Windows XP or Windows Vista, accept a license agreement, download a little code, and click the Start Test button.
It’s not a speed test (other sites do that better). Instead, the results tell you about your router and how it works with Windows. It doesn’t make any permanent changes, and it doesn’t disclose any personally identifiable information.
Recommended.
Ed Bott (@ wordpress.com) 
Very nice. Thanks for the info.
Pretty good. My router failed the two that it should have: “NAT Type” (since it doesn’t support IPv6) and “UPnP” (since it’s disabled in the router). And the program actually tells you what it’s doing, which reminded me of “Diagnose and repair” in Network Sharing Center. If I click that I get:
“The TCP setting needs to be adjusted to improve network performance.”
And the button to click was this: “Turn on tcp performance improving settings.”
Sorry, no thanks, MS. What TCP settings? Wouldn’t all settings already be tweaked by virtue of Vista being installed and it doing whatever self-tuning it does? Vista’s usually good about explaining itself.
That is indeed a test of a different Genre. Thanks Ed.
Rick,
The only reason you would get that message from Network and Sharing Center, AFAIK, is if you had used netsh int to disable autotuning. That’s a popular (and mostly bogus) tip making the rounds. Are you sure you haven’t done that?
Ed, you nailed it, since it turns out that I did even though I’d long forgotten about it. Microsoft actually forced me into it, since if you’re a heavy peer-to-peer user in Vista (and XP SP2) the Event Viewer will fill up with the infamous and annoying 4226 error–and your throughput in those apps will suffer. So I implemented the short batch file here, which in addition to patching tcpip.sys and upping the half-open connection limit, runs the command you mention. It’s infuriating that MS didn’t make this easier in Vista. Thanks.
http://www.mydigitallife.info/2007/04/09/windows-vista-tcpipsys-connection-limit-patch-for-event-id-4226/
For XP, this is a better choice: http://www.lvllord.de
Well, I finally got to use this on the laptop with Vista on it and it told me what I thought it would.
UPnP could not be run….I have it and discovery turned off.
That I could run Vista with basic networking….Been doing that since beta 1
That my old Linksys router wasn’t taking advantage of Vista.
It’s funny that it gives you a list of Vista approved routers. I’m wondering if MS is getting a kick back for that list.
Interesting tool, but there are other tools that can tell me a lot more about my network settings and ways to make them better.
Ray
Re autotuning, I happened across this today. Irony alert!
http://support.microsoft.com/default.aspx?scid=kb;en-us;940266
According to the Internet Connectivity Evaluation Tool everything works but UPnP – which is disabled in my router. I get mediocre download speeds and the internet connection on my new desktop with Vista Home Premium is a hit and miss.
Meanwhile, over on my 3-yr old laptop with XP SP2, no problems whatsoever. Excellent download speeds and no internet connection problems.
I’ve tried autotuning enabled and disabled – nothing gives me an consistent internet connection.
But I’m sure that Ed will blame everyone but MS for this fiasco. It’s amusing that he calls disabling autotuning as “mostly bogus” but MS suggests disabling autotuning if you are having problems.
Why should I have to buy a new router when my old one worked perfectly before Vista? Since when does an OS make hardware based on open standards obsolete?
Glad you’ve got your mind made up about that.
I am always interested in valid data. If you were running the same test on the same hardware with XP and Vista, it would be easy to compare. But you’re not, so we have no way of knowing where the problem is.