The MPLS WG Archive

["Ferrell, William" <William.Ferrell@titan.com>]

Sorry , I had to check my dictionary.....In laymans terms.... 

Im not saying that the proposal will not work.(this is a forum for different
ideas) I agree that checksums are used all of the time. My issues are the
overhead on the router and with the common use of checksums and its known
expoitability. If we implement another protocol on a proven fallible
technology then how does that improve the overall secure footprint of our
traffic. Again operationally I think that the solution will work , that does
not account for overt and malicious use/misues. From comprehensive
perspective this may create another wider avenue for exploitation, we should
simply consider that in the planning stages.
Will

-----Original Message-----
From: Eric Rosen [mailto:erosen@cisco.com]
Sent: Thursday, March 27, 2003 2:00 PM
To: Shahram Davari
Cc: 'curtis@fictitious.org'; 'Thomas D. Nadeau'; 'George Swallow'; W.
Mark Townsley; Andrew G. Malis; 'mpls@uu.net'; tnadeau@cisco.com
Subject: Re: [PWE3] MPLS PID 



Shahram> my proposal superior in terms of probability of false positives,

Only in  the absence of  the control word.   In the presence of  the control
word, there are no false positives. 

Shahram> Checksum is easy and is done all the time in routers

For IPv4  packets, but for non-IPv4  packets, this is extra  work that would
not otherwise need  to be done.  Extra code too (gates  or microcode), as it
would happen  in a different  forwarding path than  that used for  IPv4, and
microcoders and hardware designers are not very big on calling subroutines.