On Mon, Jan 5, 2009 at 10:14 PM, Song Haibin <[email protected]> wrote:
> Hi Bruce,
>
>>I've been in favor of adoption, and I am not going to change my opinion
> now.
>
> Many thanks!
>
>>When the revision was posted, it
>>should have been announced on the list and the changes made to it
>>brought to the attention of the wg members for discussion.
>
> I announced the new revision in the p2psip mailing list on Dec 17th.
>

You're right, my apologies.  I somehow entirely missed that.  In that
case, I really wish everyone else (such as myself) on the list had
read it and responded more quickly...

>>I have technical concerns with this draft that have been mentioned
>>before (OWD can't be measured the way this draft proposes, and I don't
>>believe the multiple-responses-for-an-echo-request technique should be
>>supported).
>
> There is an editor note in section 6.1 which indicates OWD measurement will
> need more discussion. Echo method has been limited to be used in some
> trusted overlay network where it works.
>

Measuring OWD of a multi-hop path around an overlay is entirely
possible if all the peers are synchronized to NTP servers.  It's just
that it can't be used for a single hop, and it's hard to confirm all
the peers are properly synchronized.  However, that timestamp is
useful regardless of whether it can be used for OWD.  So I would leave
the timestamp the way it is, but just add some discussion that careful
consideration needs to be given to how it's used.

While differentiating "trusted" and "untrusted" overlays is useful for
discussion points, I don't know how to apply the terms to real
networks.  Real networks are going to have a lot of shades of gray.
All you're going to wind up with is a flag in the config file saying
whether the overlay allows that multiple response echo method or not.
 However, my biggest concern is the difficulty of implementing it.
There's both a problem that it's two different ways of implementing
the same concept (which makes the implementation hard) and that this
is the only case we have for a request that gets multiple responses
each of which has meaning (much more complex than a normal
request/response).  Considering all of this, think there just needs to
be a standard one-response-at-a-time method.

Bruce


>>I don't understand why the "diagnostic server" material was added to
>>this revision.  I think it should be removed.  If something like this
>>is desired, it should be entirely separate.
>
> This section will be removed.
>
> BR
> Haibin
>
>>Bruce
>>
>>
>>
>>
>>On Fri, Dec 26, 2008 at 5:13 PM, David A. Bryan <[email protected]>
> wrote:
>>> Looking for objections, since we had consensus in the room in
> Minneapolis.
>>>
>>> Thanks for clarifying!
>>>
>>> David (as chair)
>>>
>>> On Fri, Dec 26, 2008 at 3:49 PM, Dan York <[email protected]> wrote:
>>>> David,
>>>>
>>>> Are you expecting us to send messages to the list agreeing with this?
> Or
>>>> are you taking the lack of *dissenting* messages as a sign of consensus?
>>>>
>>>> Dan
>>>>
>>>> P.S. I agree with this action, by the way, and did hum to adopt in the
> room
>>>> in Minneapolis.
>>>>
>>>> On Dec 26, 2008, at 10:08 AM, David A. Bryan wrote:
>>>>
>>>>> In Minneapolis, there was a hum taken which indicated rough consensus
>>>>> to move towards adopting the P2PSIP diagnostics draft as a working
>>>>> group item. Since there were also a number of corrections/changes
>>>>> requested, the chairs asked the authors to iterate the draft and post
>>>>> it, and then we would verify the consensus on list.
>>>>>
>>>>> The authors posted the revisions to the draft a few weeks ago:
>>>>>
>>>>> http://tools.ietf.org/id/draft-zheng-p2psip-diagnose-04.txt
>>>>>
>>>>> I'd like to ask for list consensus to verify the consensus from the
>>>>> meeting in favor of adopting this work as a WG item.
>>>>>
>>>>> David (as chair)
>>>>> _______________________________________________
>>>>> P2PSIP mailing list
>>>>> [email protected]
>>>>> https://www.ietf.org/mailman/listinfo/p2psip
>>>>
>>>> --
>>>> Dan York, CISSP, Director of Emerging Communication Technology
>>>> Office of the CTO    Voxeo Corporation     [email protected]
>>>> Phone: +1-407-455-5859  Skype: danyork  http://www.voxeo.com
>>>> Blogs: http://blogs.voxeo.com  http://www.disruptivetelephony.com
>>>>
>>>> Build voice applications based on open standards.
>>>> Find out how at http://www.voxeo.com/free
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>> _______________________________________________
>>> P2PSIP mailing list
>>> [email protected]
>>> https://www.ietf.org/mailman/listinfo/p2psip
>>>
>>_______________________________________________
>>P2PSIP mailing list
>>[email protected]
>>https://www.ietf.org/mailman/listinfo/p2psip
>
>
_______________________________________________
P2PSIP mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/p2psip

Reply via email to