RE: [bmwg] I-D ACTION:draft-ietf-bmwg-ipv6-meth-03.txt

["Chip Popoviciu \(cpopovic\)" <cpopovic at cisco.com>]

Hi David,

Your perspective on the methodology makes sense, indeed it should be
defined without assumptions of the scope for which it will be used. In
this sense there is no disagreement. The idea of the note was to put
things in context for the reader who might just take the recommendations
as they are and not relate them to his true needs. If the reader is
someone testing the maximum switch fabric performance or wants to
evaluate a limit case, that reader most likely knows what the data means
and what is its use. On the other hand, we might have readers who,
without understanding the context, would take the recommendation ad
literam.

Do you see value in providing some context to the limit case? I still
think of the comments made by Curtis (which I think are realistic):

"The ISPs don't care what BMWG specifies.  They will provide their own
number if they disagree.  The enterprise market does not have that
luxury and may also be more prone to pay attention to meaningless
benchmark figures that marketeers waive around if theirs is better."

The idea is not to discourage the use of that frame size but rather to
encourage proper interpretation of the results. The note might have been
poorly worded but that is what it intended to do. Either way, your
proposal to mention the 49 bytes option should be added.

Best Regards,
Chip  

-----Original Message-----
From: David Newman [mailto:dnewman at networktest.com] 
Sent: Tuesday, September 04, 2007 3:50 PM
To: bmwg at ietf.org
Subject: Re: [bmwg] I-D ACTION:draft-ietf-bmwg-ipv6-meth-03.txt

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On 9/4/07 11:52 AM, Al Morton wrote:
> At 10:31 AM 9/4/2007, Chip Popoviciu \(cpopovic\) wrote:
>> We have a summary of the items discussed during the Last Call and the

>> actions taken. I just sent it to Al and he will make the document 
>> available to the working group.
> 
> It's posted here:
> http://home.comcast.net/%7Eacmacm/BMWG/IPv6-Meth-Last-Call-resolution2
> .pdf
> 
> Thanks for preparing this summary!

Hi Chip et al,

Here, I hope, is my final comment on frame sizes. Characterizing
minimum-length SONET frames as being of "no practical use" is
problematic, in that practicality is relative to the user.

I know you really mean "no practical use in any conceivable production
setting" but that's beside the point in the context of bmwg docs.
Certainly for any tester asked to determine maximum switch fabric
performance, the use of minimum-length frames is not only practical but
necessary.

I'd suggest deleting this paragraph:

"Note: The 47 bytes SONET frame has no practical use since it can carry
only the 40 bytes header of an IPv6 packet and no upper layer protocol
information.  It represents however the smallest frame size for this
media type.?

If I were going to offer any comment on minimum-length SONET frames, it
would be to note that the minimum length may be 47 or 49 bytes depending
on the CRC field length in use.

dn

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (Darwin)

iD8DBQFG3bbgyPxGVjntI4IRAov7AJ0e3oCxhWvSJNUDDBX0lyzza2q1CQCg9kU/
o2tG8PNl+5RivVZftJ8o01k=
=0Y8d
-----END PGP SIGNATURE-----

_______________________________________________
bmwg mailing list
bmwg at ietf.org
https://www1.ietf.org/mailman/listinfo/bmwg

_______________________________________________
bmwg mailing list
bmwg at ietf.org
https://www1.ietf.org/mailman/listinfo/bmwg