Is int-area interested in taking this on as work?
If not, other suggestions about what to do with it?
Iljitsch
On 30 aug 2007, at 0:13, Iljitsch van Beijnum wrote:
Hi,
I've updated my multi-mtu draft:
http://www.ietf.org/internet-drafts/draft-van-beijnum-multi-mtu-01.txt
Other than some minor text issues that caught my eye I didn't change
the discussion about the larger packet downsides and CRC issues.
The router advertisement is now much simpler and contains a priority
and three MTUs: for 1000, 100/unknown-speed and 10 Mbps. However, I
do introduce two new flags: one that suggests that nodes are
"conservative" in probing neighbor MTUs (I'm not 100% convinced
about this one) and one that tells hosts do skip the probing so it's
possible to have a subnet with both nodes that support all of this
and ones that are simply configured with a non-standard MTU, i.e.,
interoperate with current jumbo frame deployments.
The switch message is gone and there is more discussion about
probing. If anyone has additional jumbo frame sizes seen in the
wild, please let me know. Probing is now done using neighbor
discovery with IPv6 but that requires breaking RFC 2461 under some
circumstances. For IPv4 it's done with jumbo ARP. The text is fairly
rough and I didn't spell check (takes forever on texts like this) so
don't mind the typos.
_______________________________________________
Int-area mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/int-area
_______________________________________________
Int-area mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/int-area
