Hi, Kiran,

I would assume that it is possible to support NVO3 deployment and
management if each overlay uses only one encapsulation protocol. The
issue is deploying a single overlay using multiple protocols - that's
where things get difficult, if not impossible. It is more challenging to
deal with heterogeneity - the OAM system would need to discover
capabilities and deploy overlays only where there was a common subset,
or match the capabilities available to the user/operator provided
constraints.

That's basically how we addressed this issue in the X-Bone overlay
deployment system. It supported different encapsulations (IPv4, IPv6,
IPv4 with IPsec, IPv6 with IPsec), but each overlay system deployed used
only one encapsulation throughout. Users/operators indicated
requirements that mapped directly to encapsulation configurations and we
had a secure capability discovery system where "invitations" asked for
particular capabilities and only nodes willing to participate responded.

I'm not sure how much of this maps to a more modern OAM system, but it
seems to me that this is an example that heterogeneous encaps protocols
can be supported usefully without too much difficulty, - again, as long
as each overlay is homogenous.

Joe



On 10/7/2016 3:29 PM, Kiran.Makhijani wrote:
>
> Anoop, Joe
>
> You are both making interesting points about capability negotiation,
> leading to this question for the group –Is it even feasible to proceed
> with OAM or control plane activity without assuming (or referring to)
> any base encap?
>
> Thanks
>
> Kiran
>
>  
>
> *From:*nvo3 [mailto:[email protected]] *On Behalf Of *Joe Touch
> *Sent:* Friday, October 07, 2016 8:37 AM
> *To:* Anoop Ghanwani <[email protected]>
> *Cc:* Sam Aldrin <[email protected]>; Bocci, Matthew (Nokia - GB)
> <[email protected]>; NVO3 <[email protected]>; Lucy yong
> <[email protected]>; Alia Atlas <[email protected]>
> *Subject:* Re: [nvo3] Discussion on encapsulation formats and next steps
>
>  
>
> If they are "ships in the night" it doesn't matter.
>
>  
>
> If they interconnect, you either have to know how to translate between
> these to sets of capabilities or fail. Using the common subset of the
> two means turning off some of the features, which avoids translation
> failures only if it can be negotiated (and both sides agree).
>
>  
>
> This is a rehash of how internetworking already deals with heterogeneity. 
>
>  
>
> Joe
>
>
> On Oct 7, 2016, at 8:13 AM, Anoop Ghanwani <[email protected]
> <mailto:[email protected]>> wrote:
>
>     Joe,
>
>      
>
>     What happens when we have encap 4 and one DC uses 10 options and
>     the other one uses on one?
>
>      
>
>     Anoop
>
>      
>
>     On Fri, Oct 7, 2016 at 7:44 AM, Joe Touch <[email protected]
>     <mailto:[email protected]>> wrote:
>
>
>
>         On 10/6/2016 9:58 AM, Anoop Ghanwani wrote:
>         > The obvious way to handle multiple DCs with different encap
>         is to use
>         > a gateway.  The simplest gateway would terminate one encap
>         domain and
>         > start another.  OAM, etc. would work within a domain.  To
>         work across
>         > domains, OAM would have to be run at a higher layer.  This
>         is not too
>         > much different than, e.g., mapping from 802.1ah/PBB to MPLS.
>         >
>         > Running multiple encaps within a single DC would be done under
>         > supervision of the NVA.  The NVA would keep track of which
>         NVEs are
>         > capable of a given encap and make sure that a given encap is
>         used only
>         > when both the source and destination NVEs can support it.
>         >
>         > It adds complexity but the problems are not insurmountable.
>
>         You are assuming that the semantics of different
>         encapsulations are
>         compatible (MTU, fragmentation, signalling, options), and
>         essentially
>         describing a translation gateway internetworking model. That
>         model was
>         replaced by a layered model using one overarching service (IP)
>         for a
>         reason.
>
>         Let's please not reinvent the failed percursors to the Internet.
>
>         Joe
>
>      
>

_______________________________________________
nvo3 mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/nvo3

Reply via email to