Scott W Brim wrote:
On Fri, Dec 06, 2002 08:15:16AM -0800, Joe Touch allegedly wrote:
>Eric Rosen wrote:
>
>>IEEE is certainly not the right place to determine how to carry
>>ethernet data and control frames over IP networks.
>
>They defined ethernet. It is they who would best determine how to
>carry ethernet over another protocol and keep current ethernet
>correctness.
To be more specific, if anything the IEEE would want to do to carry
Ethernet over IP would require modifications to protocols which the IETF
feels protective about, the work should be done under the auspices of
the IETF. They should give us requirements -- we know our protocols
better than they do.
By the same argument, they also know their protocols (L2) better than
we do.
It should be the other way around regarding protocol modification, IMHO.
Running L2 over IP is putting protocols (L2) on top of an environment
they are not designed for. The most critical ones being the latency or
timing constraints most L2 protocols have. It does not really require
modifying IP per se, because you can not get around latency in current
Internet. So one would most likely end up relaxing or changing L2
protocols to adapt to the new (and unfit for L2) environment which is
the Internet.
So if anything, IEEE should be worry more about IETF changing L2 protocols
in this case than the other way around. I personally think L2 over IP is
outside the scope of IETF, at best you can put them in Transport because
that's the result of protocol stacking. Join work with IEEE in this case
is definitely a must.
yushun.
--
____________________________________________________________________________
Yu-Shun Wang <yushunwa(_at_)isi(_dot_)edu> Information Sciences
Institute
University of Southern California