On 09/16/10 11:19 AM, Pascal Thubert (pthubert) wrote:
Hi erik,
If we compress IP in IP and a HbH header, we can hardly make that 0 bits
overhead. The FL allows a total cost of 20 bits, whereas the
compression will end up with 5 or 6 octets minimum...
You misunderstood my suggestion.
Erik
Pascal
-----Original Message-----
From: Erik Nordmark [mailto:erik.nordm...@oracle.com]
Sent: Thursday, September 16, 2010 6:44 PM
To: Pascal Thubert (pthubert)
Cc: Philip Levis; Carsten Bormann; r...@ietf.org; ipv6@ietf.org;
m...@sandelman.ca; Brian E Carpenter
Subject: Re: [Roll] Flow Label: 12 bits mutable and 8 bits immutable
On 09/16/10 05:07 AM, Pascal Thubert (pthubert) wrote:
Has anybody discussed adding a header with just the 3 bytes you
need
*before* the IP header?
We've discussed using forms of tag switching from the beginning of
the
WG but there must be an IP based reference model before we start
doing
the tricks.
You can have a reference model which uses IP-in-IP plus a destination
options
to carry the 3 bytes, and then have an optimization for low speed
networks like
6lowpan that sticks those bytes somewhere else as Carsten suggested.
Architecturally, in both cases, we'd end up with the 3 bytes before
the
(inner) IP header.
Erik
--------------------------------------------------------------------
IETF IPv6 working group mailing list
ipv6@ietf.org
Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6
--------------------------------------------------------------------
