Re: [nmrg] Discovery of resources in NFV environments

[Carlos Jesús Bernardos Cano <[e-mail]>]

Hi Brian,

On Mon, 2018-10-22 at 13:43 +1300, Brian E Carpenter wrote:
> Hi,
> 
> What is the argument for using a low-level mechanism such as ND,
> which requires modifications to the basic IPv6 stack?
> 
> We have other general discovery mechanisms, such as GRASP for
> autonomic scenarios and DNS-SD for more conventional scenarios
> (and those can be blended if necessary, draft-eckert-anima-grasp-
> dnssd)
> which don't require low-level stack updates. That would have the
> advantage
> of more flexibility for the option formats, instead of the rigidity
> required by RFC4861.

We are exploring different options. So far we have looked into ND
extensions as a fast way of conveying some (but not all) the info about
the resources a node has available. But we want to also explore other
ways, like the ones you mentioned, and also combinations (like first
knowing if a neighbor has certain capabilities and then if required to
request more detailed information in a bilateral dialogue using a more
conventional discovery mechanism on top of IP).

Another reason for considering extensions to ND is that the scenarios
we have in scope are very dynamic and we want to get knowledge of
resources appearing and disappearing as fast as possible.

In any case, we are just exploring options. We'd like to first get
comments on the actual problem we are addressing (enabling dynamic
advertisement, discovery and association of resources and orchestrators
in a virtualization environment), and then look into potential
solutions.

Thanks for your comments.

Carlos

> 
> Regards
>    Brian
> 
> On 2018-10-22 04:22, Carlos Jesús Bernardos Cano wrote:
> > Hi,
> > 
> > We are working on a draft that describes mechanisms allowing
> > dynamic
> > discovery of virtualization resources and orchestrators in IPv6-
> > based
> > networks. The problem of discovery and advertisement of
> > functions/resources is somehow generic in network management,
> > though
> > for the case of NFV we believe it is more relevant because of the
> > dynamic nature of some NFV deployment scenarios (such as the so-
> > called
> > fog computing one).
> > 
> > We started a discussion in the NFVRG, where we presented the
> > problem
> > and our ideas a couple of times with good feedback. Since the NFVRG
> > is
> > closed, we'd like to continue getting feedback and potential
> > interest
> > in working on this ideas in other WGs/RGs. We believe NMRG is a
> > good
> > match, and therefore we'd like to ask for some comments, and
> > eventually
> > a slot for presentation in Bangkok if possible.
> > 
> > Our document is the following:
> > 
> > "IPv6-based discovery and association of Virtualization
> > Infrastructure
> >  Manager (VIM) and Network Function Virtualization Orchestrator
> > (NFVO)"
> > 
> > draft-bernardos-intarea-vim-discovery-00
> > 
> > 
https://datatracker.ietf.org/doc/draft-bernardos-intarea-vim-discovery/
> > 
> > Looking forward for your comments!
> > 
> > Thanks,
> > 
> > Carlos
> > 
> > _______________________________________________
> > nmrg mailing list
> > [e-mail]
> > https://www.irtf.org/mailman/listinfo/nmrg
> > 
> 
> _______________________________________________
> nmrg mailing list
> [e-mail]
> https://www.irtf.org/mailman/listinfo/nmrg