>If the delay wasn't so precise (12 seconds for every hit), I might agree. I have seen this enough in past months that I thought it might be an internal (UT) problem so I mentioned it. As no one else seems to have seen the problem, I have to assume somewhere else along the route was introducing the delay. If (when) I see it again, I'll give you a shout if you are online. I'll also get some ping data for the route.

If it would have been internal, we wouldn't be able to process 4 hits per seconds during rush hour. If you do a small calculation, that would be 48 hits before yours kicks in. So, a rapid long line would completely jam the server after a few seconds. There is definitely something in the route you are using which is failing. Usual test in such circumstances is to hook to the Internet from another point of entry, such as dialup, and see if you could get better results.