[4161] in linux-net channel archive
Re: SYN floods
daemon@ATHENA.MIT.EDU (Ralf Baechle)
Fri Aug 23 05:49:18 1996
To: submit-linux-dev-net@ratatosk.yggdrasil.com
From: ralf@julia.de (Ralf Baechle)
Date: 23 Aug 1996 08:06:36 GMT
In article <Pine.LNX.3.91.960822001606.133F-100000@cirrus.bluesky.net>, Racer X <shagboy@wspice.com> writes:
|> On 20 Aug 1996 nelson@crynwr.com wrote:
|> I can STILL say I come from any IP I want. Your provider will (should)
|> have a route to the class C I am connected by. I can (at the very least)
|> say I am coming from any of the 254 possible hosts on that net.
|>
|> Besides that, how does a backbone router ultimately know "where" a packet
|> came from? I guess maybe it could check to see which side it comes in,
|> but that's an awful big table it has to keep in memory. Considering how
|> often many major routers go down already, I'd really rather do it
|> myself...
You can do the filtering required assuming symmetric routing with no extra
memory. For assymetric routing this filtering is very difficult because
a router knows where to route to but it doesn't necessarily know through
which interface traffic from which net comes. Even worse, each router has
to the checks or manipulated traffic could passed on in a way that the
destination site or another router on the way cannot recognize.
Unfortunately assymetric routing is something that does exist in the
real world, not only in theory.
Ralf
--
A weird imagination is most useful to gain full advantage
of all the features - manpage of amd(8).
