If you really want to test raw performance, why not just enable the chargen service on a router in the core and let him suck that down and measure throughput. No more effective DOS attach than 4 simultaneous chargen sessions on a T1. I suspect a 7513 should be able to spit a few characters per second :) Eric At 11:24 AM 7/9/97 -0700, John Leong wrote:

n Mon, 7 Jul 1997, Jack Rickard wrote:

...

I discussed the universal test page

...

suggestion with Gene Shklar this afternoon and we will make it so.

Universal test page is certainly a [good] step - but it is a step (one among many) towards [comparative] testing the performance between the test machine and a *SPECIFIC* web server .... but not the peformance of the backbone (even though it clearly is a major component of the result). There are simply too many uncontrolled variables still outstanding.

By the way, we have given considerable thought on the attributes of such standardised test page. ... starting with what is the test about. It would be a good idea if a sub group of interested parties can get off line and work on this under some frame work (IETF, NANOG, CAIDA, ... or just an ad hoc BOF group) and then report the finding back to the mailing list.

So, if anyone is interested to work on a standardized web page, send me mail and I will try to get a seperate discussion list going, do some serious work ... and reduces the S/N ratio of this list.

Regards, John Leong