I agree with James. Virtualization is getting intense interest in
industry but in a way that is very different from the way projects like
GENI and FEDERICA are using it. Operators use virtualization as a way
of bringing up and managing services. GENI uses virtualization to allow
researchers to share deeply-programmable experimental infrastructure.
The constraints and goals are quite different, I think. How are these
aspects related? Can an RG be used to bring interested parties from
both communities together? What can they learn from each other? It
might be worth trying to chat with some vendors (HP? Cisco?) and
operators (MVNOs?) before trying to answer these questions. Since,
AFAICT, they have been largely absent from the discussion so far.
Off the top of my head, here are a few examples of challenges GENI faces
related to virtualization:
* How is isolation between slices handled? To strict a requirement
and you loose flexibility, too loose and experiments are
influenced by other slices.
* How is repeatability handled? The virtual world is dependent on
the real infrastructure beneath it. How much should it be
measured/controlled to permit repeatable experiments?
* How is visibility handled? Users may be interested in fine grain
information about the real devices in the path. By definition,
this is hidden from users via virtualization. How is visibility
controlled?
* How do the three topics above (isolation, repeatability, and
visibility) impact experiment design and resource discovery?
* Can slices with very diverse goals in the above topics share
infrastructure?
There are probably others but I'd need more coffee to produce them. :)
cheers,
--aaron
James Kempf wrote:
> The charter is currently a bit fuzzy. I don't see much value in yet
> another forum to discuss architecture or other theoretical concerns,
> except perhaps to keep contact with important academic work in this
> area.
>
> If there is any reason at all for an IETF-affiliated group on
> virtualization, it has to do with interoperability, IMHO. Currently,
> there is much work going on in GENI on control frameworks for
> provisioning a virtual aggregate (a network slice and compute
> resources). AFICT, this work is primarily directed at the US academic
> institutions and the operators involved are primarily nonprofits
> (Internet 2 and the nonprofit regionals) due to the US funding model
> (through NSF). I don't know what is going on in Europe, Japan, or China
> with respect to prototype frameworks for virtual network control, though
> I am somewhat familiar with virtualization architectural work in Europe.
> The danger is that we end up in 5 years time with national or regional
> control frameworks for virtualized resource allocation that don't
> properly interoperate.
>
> I think the RG should focus on bringing together researchers in
> different geographic areas to define interoperability between control
> frameworks, and should include both nonprofit operators and for-profit
> operators, vendors, and researchers. Ideally, in perhaps a year or two,
> the issues should be well enough sorted out for an IETF WG to form that
> can standardize an interoperability framework.
>
> There may be some other topics that an RG could consider, but I think it
> would be better to have a specific focus that would serve IETF well.
>
> jak
>
>
>> -----Original Message-----
>> From: nvrg-bounces at listserv.gwdg.de
>> [mailto:nvrg-bounces at listserv.gwdg.de] On Behalf Of Martin Stiemerling
>> Sent: Friday, May 22, 2009 12:34 AM
>> To: nvrg at listserv.gwdg.de
>> Subject: [nvrg-bof] Updated charter proposal
>>
>> Hi all,
>>
>> Here is the updated charter proposal. Let me know your comments.
>>
>> Martin
>>
>> Draft charter for the Virtual Networks Research Group VNRG) proposal
>>
>> VERSION: 2009-05-22 0800 UCT
>>
>> Recent developments in networking are aiming at better
>> utilizing infrastructure in terms of reusing a single
>> physical or logical resource for multiple other network
>> instances, or to aggregate multiple these resources to obtain
>> more "functionality". These resource can be network
>> components, such as, for example, routers, switches or hosts,
>> and also services, such as, for example, name mapping
>> systems. Typically, this is referred to as Virtual Networks,
>> where a resource is either re-used for multiple networks or
>> multiple resources are aggregated for virtual resource.
>> Important properties of Virtual Networks are i) the level of
>> participation of each resource and ii) the clear separation
>> of any virtual network to all others. Each resource can be
>> sliced so that it can be part of multiple Virtual Networks,
>> but on the other hand does virtualization guarentee the clear
>> separation of each network, so that actions in one network do
>> not affect the operation of any other network.
>>
>> However, in the network community, Virtual Networks is a very
>> broad term, ranging from running multiple wave lengths over a
>> fiber, MPLS, virtual routers, to overlay systems. This leads
>> to deployment of single technologies in parts of the Internet
>> or other IP-based networks, but lacks a common understanding
>> of what virtualized networks is causing to IP-based networks,
>> or how Virtual Networks is applied in favorable way. This
>> leads to the introduction of virtualization in an
>> uncoordinated way between the various players, such as
>> network operators, vendors, service providers and testbed
>> providers (e.g. GENI, FEDERICA, etc) without considerations
>> about the overall impact on the system level.
>>
>> The Virtual Networks Research Group (VNRG) provides a forum
>> for interchange of ideas among a group of network researchers
>> with an interest in network virtualization in the context of
>> the Interent and also beyond the current Internet.
>>
>> The RG works on a set of principles of virtual networks that
>> a single physical resource can be re-used by multiple
>> entities with a clear separation between the actions taken by
>> the single entities. Virtualization delivers an abstraction
>> to the user (i.e., not necessarily a human but a service or
>> whatever) referring to the decoupling from the physical
>> resource, i.e., the abstraction is not bound to a single
>> resource but can be relocated but is a logical structure.
>> Virtualization also offers recursion, i.e., an already
>> virtualized network can again include virtual networks.
>>
>>
>> The group will address the following research challenges:
>> - Consider a whole system for virtualized networks and not only single
>> components or a limited set of components;
>> - Identifying architectural challenges resulting from virtual
>> networks;
>> - Recursive network management of virtual networks;
>> - Emerging technological and implementation issues.
>>
>>
>> Web site: http://user.informatik.uni-goettingen.de/~stiemer/nvrg/
>>
>> Mailing list: Open mailing list to anybody, just needs signing up
>> with the list
>> list's address: nvrg at listserv.gwdg.de
>> signing up: https://listserv.gwdg.de/mailman/listinfo/nvrg
>>
>>
>>
>>
>> stiemerling at nw.neclab.eu
>>
>> NEC Laboratories Europe - Network Research Division
>> NEC Europe Limited | Registered Office: NEC House, 1 Victoria
>> Road, London W3 6BL | Registered in England 2832014
>>
>>
>>
Note Well: Messages sent to this mailing list are the opinions of the senders and do not imply endorsement by the IETF.