Re: sequence number wrap

["Steven M. Bellovin" <smb@research.att.com>]

In message <200304241721.h3OHLMtD019688@rotala.raleigh.ibm.com>, Thomas Narten 
writes:
>In two separate documents I've looked at this week, I've seen protocol
>support for detecting out-of-sequence packets that uses a rather small
>sequence number space.
>
>draft-ietf-pwe3-ethernet-encap-02.txt uses a 16-bit field (this is to
>ensure that Ethernet frames don't get reordered across IP or MPLS).
>
>draft-ietf-l2tpext-l2tp-base-07.txt uses 24 bits. (this is for when
>carrying packets other than PPP over l2tp).
>
>Am I the only person who worries that these values are way too small
>given today and tomorrow's networking speeds?
>

It's certainly worth attention.  I pushed back on a MIB that had 32-bit 
counters, for similar reasons. 

		--Steve Bellovin, http://www.research.att.com/~smb (me)
		http://www.wilyhacker.com (2nd edition of "Firewalls" book)