Re: [nvo3] Fwd: DRAFT Charter Update for Discussion

[Tom Herbert <[e-mail]>]

On Thu, Aug 14, 2014 at 10:45 AM, Benson Schliesser
<[e-mail]> wrote:
> Hi, Tom. Thanks for the comments.
>
> Just to reiterate / add to Matthew's response: The architecture meant to
> describe how the control and data plane protocols relate, not apply to the
> datacenter in some broader sense. Maybe it's reasonable to think of this in
> terms similar to how IP addressing is coherent across both the IP header and
> routing protocols... In the overlay DCVPN we have some kind of mapping
> system that ties the planes together, and this would be described by the
> architecture.
>
> That being said, assuming this description makes sense to you, do you have
> any specific suggestions for improving the charter text? If our intention
> wasn't made clear by the charter that we proposed then, by all means, we're
> open to improvements.
>
I guess my real question was: will the output of nvo3 be a proposed
set of standard data and/or control plane protocols for network
virtualization. Assuming that, I like Don's suggested wording in
describing DCVPN as "a set of protocols and/or protocol extensions."

Thanks,
Tom

> Thanks,
> -Benson
>
>
>
>
> On Thu, Aug 14, 2014 at 3:07 AM, Bocci, Matthew (Matthew)
> <[e-mail]> wrote:
>>
>> Tom
>>
>> Thanks for your comments.
>>
>> The intention is not to try to impose an architecture on data centers. The
>> architecture really provides a context in which the data and control plane
>> protocols operate, and helps to define the scope of what we can define, or
>> what new protocols we need to define in NVO3. If you look at other Wgs
>> defining VPN-like technologies (L3VPN, L3VPN, PWE3, etc), they all have
>> frameworks or architectures that do this. Ideally the architecture should
>> actually help in describing modularity, in that different control/data
>> plane protocols could be slotted in to provide different capabilities, as
>> required.
>>
>> Regards
>>
>> Matthew
>>
>> On 13/08/2014 21:31, "Tom Herbert" <[e-mail]> wrote:
>>
>> >On Wed, Aug 13, 2014 at 12:41 PM, Benson Schliesser
>> ><[e-mail]> wrote:
>> >> Just a reminder that Matthew and I are looking for feedback on the
>> >> draft
>> >> text of a new NVO3 charter. Can it be that we wrote a charter so
>> >> perfect
>> >> there are no comments..?
>> >>
>> >Maybe just a request for clarification...
>> >
>> >Unless I am misreading this, it seems to me that the intent is to
>> >create a packaged solution for data centers that combines data plane,
>> >control, and architecture. I think these are actually very separate
>> >facets and should really be mostly separate tracks. I would liken this
>> >to how IP, routing protocols, and our current data center architecture
>> >has evolved. These were never defined together and so the dependencies
>> >between them have always been quite minimal. This has allowed us to
>> >completely change one part with redoing the rest of world (e.g.
>> >IPv4->IPv6, decentralized routing to open flow,..).
>> >
>> >It might also be nice to clarify a little more what the ultimate
>> >output is of the WG. I assume the tangible output here would be
>> >standardized data plane protocol(s) and control plane protocol(s).
>> >Architecture is nice to provide a context for the protocols, but not
>> >really something that can be standardized in itself. Every existing DC
>> >already has an existing architecture (even before virtualization), it
>> >is much more likely that we'd want to adapt and leverage the existing
>> >architecture as much as possible rather than change the whole world to
>> >accommodate NV-- which I guess is another way of saying data plane,
>> >control plane, architecture need to be considered independently.
>> >
>> >Hope this helps :-)
>> >
>> >Tom
>> >
>> >> The draft charter text is quoted in my email below. For reference it
>> >> can
>> >> also be found at
>> >>
>>
>> >> >>http://svn.tools.ietf.org/svn/wg/nvo3/charter-ietf-nvo3-01-rev-20140808.t
>> >>xt
>> >>
>> >> Cheers,
>> >> -Benson & Matthew
>> >>
>> >>
>> >>
>> >> ---------- Forwarded message ----------
>> >> From: Benson Schliesser <[e-mail]>
>> >> Date: Fri, Aug 8, 2014 at 4:53 PM
>> >> Subject: DRAFT Charter Update for Discussion
>> >> To: [e-mail]
>> >>
>> >>
>> >> Dear NVO3 Contributors -
>> >>
>> >> As discussed during the NVO3 meeting at IETF-90 in Toronto, the chairs
>> >>have
>> >> been drafting a new / revised charter for the WG. The latest draft
>> >>charter
>> >> text is below for your review.
>> >>
>> >> Frankly, we suspect that some of the wording could be further improved
>> >>to be
>> >> more clear and/or precise. With that in mind we ask for your help -
>> >>please
>> >> let us know if something seems unclear or if you have suggestions for
>> >>better
>> >> wording.
>> >>
>> >> If you have feedback please post it to the list for discussion within
>> >>the
>> >> next 2 weeks.
>> >>
>> >> Thanks,
>> >> -Benson & Matthew
>> >>
>> >> An NVO3 solution, also known as a Data Center Virtual Private Network
>> >> (DCVPN),
>> >> is a set of protocols and/or protocol extensions that address the
>> >> issues
>> >> described by draft-ietf-nvo3-overlay-problem-statement consistent with
>> >>the
>> >> approach described by draft-ietf-nvo3-framework.
>> >>
>> >> NVO3 will document DCVPN requirements for both control plane
>> >>protocol(s) and
>> >> data plane encapsulation format(s), as well as management, operational,
>> >> maintenance, troubleshooting, security and OAM protocol requirements.
>> >> Additionally, NVO3 will document common use-cases for DCVPN solutions.
>> >>
>> >> Consistent with the documents described above, the NVO3 WG will
>> >>document an
>> >> architecture for DCVPNs within a data center environment based on the
>> >> following
>> >> architectural design points:
>> >> -   A logically centralized NVA control plane
>> >> -   Support for an underlay IP data plane between NVEs
>> >>
>> >> Based on this architecture the NVO3 WG will develop one or more NVO3
>> >> solutions.
>> >> This may include documenting applicability of existing protocols,
>> >> contributing
>> >> to the development of protocol extensions by other WGs and/or SDOs,
>> >>and/or
>> >> developing new protocols as appropriate.
>> >>
>> >> Solutions and/or protocols that were developed outside of the IETF, but
>> >>not
>> >> developed by another SDO, and that have multiple interoperable
>> >> implementations
>> >> may be adopted by the NVO3 WG for further development, based on WG
>> >> consensus,
>> >> if requested by the authors .
>> >>
>> >> If the NVO3 WG anticipates the adoption of the technologies of another
>> >>SDO,
>> >> such as the IEEE, as part of any DCVPN solution, it will liaise with
>> >>that
>> >> SDO
>> >> to ensure the compatibility of the approach.
>> >>
>> >>
>> >>
>> >>
>> >>
>> >> _______________________________________________
>> >> nvo3 mailing list
>> >> [e-mail]
>> >> https://www.ietf.org/mailman/listinfo/nvo3
>> >>
>> >
>> >_______________________________________________
>> >nvo3 mailing list
>> >[e-mail]
>> >https://www.ietf.org/mailman/listinfo/nvo3
>>
>