help required

[Brad Knowles]

At 5:41 AM +0200 9/22/01, Terrence Koeman wrote:

>  Well thank you for your kind words on the fact that I'm offering my
>  bandwidth and servers for free to anyone who wants to use them.

	Hey, you're the one offering up machines capable of being cache 
poisoned as a free service to the Internet.

>  The servers are in fact a cluster of 5 machines and handle about 50
>  million queries a day.

	Right.  Let's do the math:

		50,000,000 / 86400 = 578.7037

			578.7037 / 5 = 115.74074

	Sounds pretty close to the numbers I calculated.

>                          I'm wondering how you have 'benchmarked' my
>  servers?

	There's a program called "queryperf" that ships with the latest 
betas and release candidates of BIND 9.2.0.  Indeed, I'm in the 
process of working up a wrapper script for this program to automate 
the stress testing/benchmarking of one or more servers for a 
particular zone.

>            I think that you have measured nothing more than the latency,
>  and I highly recommend 'ping' or 'traceroute' for that.

	That's pretty hard to do if you block both ping & traceroute at 
your firewalls, or if your upstream provider (Level 3) blocks ping & 
traceroute.

-- 
Brad Knowles, <brad.knowles at skynet.be>

H4sICIFgXzsCA2RtYS1zaWcAPVHLbsMwDDvXX0H0kkvbfxiwVw8FCmzAzqqj1F4dy7CdBfn7
Kc6wmyGRFEnvvxiWQoCvqI7RSWTcfGXQNqCUAnfIU+AT8OZ/GCNjRVlH0bKpguJkxiITZqes
MxwpSucyDJzXxQEUe/ihgXqJXUXwD9ajB6NHonLmNrUSK9nacHQnH097szO74xFXqtlbT3il
wMsBz5cnfCR5cEmci0Rj9u/jqBbPeES1I4PeFBXPUIT1XDSOuutFXylzrQvGyboWstCoQZyP
dxX4dLx0eauFe1x9puhoi0Ao1omEJo+BZ6XLVNaVpWiKekxN0VK2VMpmAy+Bk7ZV4SO+p1L/
uErNRS/qH2iFU+iNOtbcmVt9N16lfF7tLv9FXNj8AiyNcOi1AQAA