Hi Paul,

We are all in agreement here.

The best approach is to not get into specific details (I agree that the orginal 
text was misleading - even if it was not the intent).

Hopefully, the newly suggested introduction will be satisfactory to everybody.

Marc

________________________________
From: Paul Unbehagen [mailto:[email protected]]
Sent: Friday, June 29, 2012 11:18 PM
To: AshwoodsmithPeter
Cc: Eric Gray; LASSERRE, MARC (MARC); [email protected]
Subject: Re: [nvo3] call for adoption: draft-lasserre-nvo3-framework-02

I have personally seen numerous DC's deployed over the last year that have used 
SPB to create an end to end Ethernet underlay to solve many of the problems 
that nvo3 is also trying to fix with overlays.   VM's are moving across cities 
transparently to the IP layer above in both single and multi-tenant 
environments.

Obviously some customers may want an overlay for other reasons, but Janos' 
points are correct.

Marc, perhaps if the text on Ethernet capabilities was updated to reflect what 
is possible today, it will clear up the confusion.
--
Paul Unbehagen

Sent from my iPad


On Jun 29, 2012, at 2:33 PM, AshwoodsmithPeter 
<[email protected]<mailto:[email protected]>> wrote:

Agreed, I think that stating that those limitiations apply to all "Existing 
virtual network models used for data center networks" is perhaps a bit sweeping 
and could use a bit of qualification otherwise its untrue.
Peter
________________________________
From: [email protected]<mailto:[email protected]> 
[mailto:[email protected]] On Behalf Of Eric Gray
Sent: Friday, June 29, 2012 3:43 PM
To: Marc Lasserre
Cc: [email protected]<mailto:[email protected]>
Subject: Re: [nvo3] call for adoption: draft-lasserre-nvo3-framework-02
Marc,
    Let's not try to make something hard out of this.
    Janos suggested that certain statements made in your draft are not 
particularly
true (or applicable) if existing standard capabilties have been deployed.
    You responded (paraphrasing) that - while this is true - there are networks 
where
these capabilities are not part of the layer-2 deployment, and there are 
customers
who would prefer a layer-3 solution.
    So far, I have no gripe with this.
    However, Joel points out that the current wording makes it appear that the 
driving
justification for doing this work in layr-3 is that existing layer-2 
standards/solutions
won't work.  He further points out that this is not consistent with your own 
admission
that there are well-known layer-2 standards/solutions that address the 
limitations you
list.
    Apparently, it was not your intent to make it sound this way.
    Without going line by line through the introduction, the general changes 
required
to bring what you are saying back into alignment with what we all apparently 
agree
is the case are relatively obvious and simple.
1) The simplest change would be to remove the statements about layer-2 
limitations
    as Janos has suggested.  I think we all understand why perhaps this is not 
the
    way you want to go.  Fine.
2) If you are going to list the limitations that you now list, you need to make 
it clear
    these are "perceived" limitations based on currently deployed layer-2 
networking
    technology, in some networks.
3) If one of the key reasons why we want to define a layer-3 solution is that 
there is
    a demand from customers for a layer-3 solution, simply say that rather than 
giving
    inaccurate second-hand information about "perceived" reasons for this 
preference
    based on out-of-date layer-2 deployments.
    Does this help?
--
Eric
_______________________________________________
nvo3 mailing list
[email protected]<mailto:[email protected]>
https://www.ietf.org/mailman/listinfo/nvo3
_______________________________________________
nvo3 mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/nvo3

Reply via email to