nanog mailing list archives

Re: UUNet Offer New Protection Against DDoS


From: "Paul G" <paul () rusko us>
Date: Wed, 3 Mar 2004 04:04:08 -0500


erik,

----- Original Message ----- 
From: "Erik Haagsman" <erik () we-dare net>
To: "Paul G" <paul () rusko us>
Cc: "Deepak Jain" <deepak () ai net>; "william(at)elan.net" <william () elan net>;
"John Obi" <dalnetuzer () yahoo com>; <nanog () merit edu>
Sent: Wednesday, March 03, 2004 3:47 AM
Subject: Re: UUNet Offer New Protection Against DDoS



On Wed, 2004-03-03 at 09:26, Paul G wrote:
cant speak for them, but this would be my preferred first step. next
step
is, of course, an attempt to filter on {source, unique characteristics,
what
have you} and removing the blackhole.

What most people seem to forget is that neither of these steps actually
counter the DoS...they merely make the DoS as invisible as possible to
customers

correct. from our pov, it is gone. given that 'solving the problem' is not
always possible, this is almost as good as it gets in the real world.

while the traffic keeps hitting the carrier in question. For
the large carriers this is only a minor inconvenience.
For smaller carriers or for co-location facilities/NSP's that are
relying on not-so-clueful carriers (read: carriers not supporting any
kind of communities with possible lack of pro-active network management
and/or bad communications) this is a BIG problem. Even though they might
take the heat off the targeted customer, they could be in for a rough
ride themselves as the DoS keeps going and going.

we tend to get small ddos (a few hundred megs) that are more of an annoyance
than anything else, at least before they hit the customer-in-question 's
faste handoff.

I haven't seen any major press-releases on actually solving the problem
instead of hiding it... (granted...I haven't put out one either :-)

<grin>. in other news, noone has solved the perpetuum mobile problem either.
as a carrier, your job is to solve the problem for the customer. this
includes staying up afterwards.

paul


Current thread: