Re: [nvrg-bof] poll on charter focus - RESPONSES DUE OCT 23

[James Kempf <james.kempf at ericsson.com>]

Hi Thomas,

I'm inclined to agree with your view below.

In particular, my understanding is the IRTF is supposed to work on topics that are pre-standardization but at some point will transition to IETF for standardization. Not on general networking (or even broader, systems) research.

Now, I have a fairly narrow view of what topics are appropriate for successful standardization (by which I mean that the work is not just standardized but actually deployed). This view comes from having spent something like 10 years in IETF, mostly doing interesting technical work that has seen little deployment. When I look at cloud computing and network virtualization, as you say, there is a lot of activity going on in this area, including in industry forums that are doing "standardized" APIs and such for cloud. Most of this activity is around systems that *are* actually getting deployed. I have a difficult time seeing where IETF could contribute.

One area that comes to mind, which I brought up about a year back, is a standardized control framework for service deployment on virtualized systems. GENI is working on 5 (I think) such frameworks, all based on Web services. I believe their plan is to at some point select one or a collection of these. There are also a couple of FP7 EU projects currently being planned that have control frameworks as part of their work. So I could see us ending up somewhere down the line with multiple such frameworks, and an RG could be a place to do an assessment and make a recommendation to IETF about what to standardize. The important point here is that the work on actually developing the frameworks is going on outside IETF.

However, I don't think that this topic is enough to form an RG. It could be handled by a WG at the time the need for a standard became apparent.

Another topic is "over the top" v.s. "close to the metal" network virtualization. Talking to GENI researchers, they seem to be planning to run virtualized networks as overlays like X-Bone was. This is in contrast to the, shall we say, "original promise" of virtualized networks, which is that they provide isolation between traffic on different "slices". Providing isolation while still providing the cost benefits of statistical multiplexing, especially across service providers, is a challenging research problem. Most L2 VPNs with service guarantees are expensive and time consuming to provision and confined to a single provider, which is probably why academic researchers aren't planning on using them. And statistical multiplexing is not possible because the VPN customers ping constantly to make sure they are getting their SLAs fufilled.

This is the kind of topic that could result in some useful output from an RG. It would require substantial service provider participation with a commitment to actually deploying prototypes, academic participation to develop the protocols, and of course vendor participation to actually develop the prototypes. I don't know whether the I*TF is today capable of this kind of thing, though, but I don't see any other context in the US where this kind of co-operation between interest groups is possible (in Europe, the big EU projects regularly have co-operation of this kind).

Anyway, so absent a commitment for that kind of co-operation, I'd say that IETF meetings don't need another RG for people to come and present work that they could do at an academic conference or another industry forum. 

Maybe I'm missing something, but most of the other proposals I've seen on this list haven't been particularly compelling (for example, I don't think arguing about what to call things, then writing a terminiology draft is particularly useful).

			jak



 

> -----Original Message-----
> From: nvrg-bounces at listserv.gwdg.de 
> [mailto:nvrg-bounces at listserv.gwdg.de] On Behalf Of Thomas Narten
> Sent: Monday, October 12, 2009 6:17 AM
> To: Joe Touch
> Cc: nvrg at listserv.gwdg.de
> Subject: Re: [nvrg-bof] poll on charter focus - RESPONSES DUE OCT 23
> 
> Let me be a bit of a naysayer here about this entire effort. 
> This list was set up more than a year ago, and still seems to 
> be in the earliest stages of figuring out what, if anything 
> to do. That is not a good sign...
> 
> Meanwhile, virtualization (and cloud computing) continues to 
> be a very hot topic in many places, including:
> 
>   - DMTF (OVF, Virtualization Management Forum)
>   - OGF Open Cloud Computing Interface WG (OCCI-WG)
> 
> Not to mention all the work various vendors are doing.
> 
> It seems to me like this group is pretty detached from all 
> the related work that is going on, and based on the pace of 
> progress so far, doesn't really have a critical mass to do anything.
> 
> What am I missing?
> 
> (I'm not necessarily opposed to a WG being formed, but I 
> don't see the point in bothering if the group isn't actually 
> going to produce anything useful...)
> 
> Thomas
>