[e2e] IP options inserted in transit
Angelo Dell'Aera
buffer at antifork.org
Sat Aug 9 09:23:39 PDT 2003
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On Sat, 9 Aug 2003 11:34:03 -0400
RJ Atkinson <rja at extremenetworks.com> wrote:
>
>On Friday, Aug 8, 2003, at 19:40 America/Montreal, Angelo Dell'Aera
>wrote:
>> I don't agree. Letting intermediate routers adding/deleting IP options
>> on-the-fly means just losing performance. If a router is asked to
>> modify every IP header (whose size could be variable), it will waste a
>> lot of time searching where it's possible to add one or more bytes or
>> which bytes to remove.
>
>This is untrue, particularly for well-designed routers.
>
>So far, I don't see any reason to add such an IP option on the fly,
>but it is not hard to do in a well-designed ASIC-based router.
>One does have to have very good Verilog designers, of course.
I agree that it's possible to design routers in such a way as to do it
but if the router is called to parse every packet searching for
on-the-fly added IP options (and this means it is called to parse what
is beyond the IP header searching for "something useful"), this will
cause a loss of performance. Maybe, it will be a little loss but you
should agree with me that the router will need more time for the single
packet processing.
Moreover, I completely agree with you when you say you find no reason
for doing it.
Regards.
- --
Angelo Dell'Aera 'buffer'
Antifork Research, Inc. http://buffer.antifork.org
PGP information in e-mail header
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2 (GNU/Linux)
iD8DBQE/NSALpONIzxnBXKIRAnXrAKDILkMNciJy4Uq2i6pno0Vb1Es2gQCgn5Ri
tK2+l9jMx6RBeISjYofJvZQ=
=QNOZ
-----END PGP SIGNATURE-----
More information about the end2end-interest
mailing list