[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [vnrg] way forward on VNRG definitions



Dear Martin, all,

- Inter-domain-wide operation across multiple administrative domains.
 Comparable to the issues in QoS research we may be able to easily
 provide dedicated services to users within our domain, but providing
 these services inter-domain-wide may have some technical or business
constraints.
I tend to disagree with this point.
The abstraction as part of the virtualization should hide whether resources belong to one or multiple domain. Scenario I am thinking of: global network service provider X and global network service provider Y obtain guaranteed virtual resources from virtual infrastructure provider P who aggregates and VIRTUALIZES resources (probably also partly virtual resources) obtained from local infrastructure provider A and in another local infrastructure provider B. Those local infrastructure providers might also deliver virtual resources directly to local service providers xa, xb, ya and yb.

Having said that, doing is much harder than saying a couple of words. Still I believe the abstraction and aggregation features of virtualization should enable overcoming multi-domain issues. You are right, business constraints can probably not be solved by virtualization (probably not by any technical solution...). Having said that: what is the business scenario in which a provider delivers virtual network instances but putting in place business constraints preventing extensions to other domains?

Kind regards,

Didier

Martin Röhricht wrote:
Dear Joe,

On 08.06.2010 18:57 Joe Touch wrote:
As a starting point, we would like to have everyone on the list answer the following questions. We would prefer short text rather than full paragraphs
extracted from any I-D or other document.
-----------------------------------------------------

Starting questions:

1. how do you define VNs?

A VN is a network that appears to its users as an ordinary network by providing ordinary interfaces and services but which is actually decoupled from the physical substrate.

    1.a. what are the key components?

Virtual nodes, virtual links, management entities

    1.b. what is the relationship between these components?

An arbitrary number of virtual links may be spanned between an arbitrary number of virtual nodes. A physical node may be equipped with multiple virtual nodes. Virtual links are controlled by management entities residing on physical nodes and acting as mediators between pyhsical and virtual nodes.

    1.c. what is the characteristic behavior/capability of the
    resulting system?

The resulting virtual network provides its users the illusion of using a dedicated (physical) network with a set of guaranteed resources. It should provide its services isolated from other virtual or physical networks that actually run on the same hardware. Network management operations should be performed transparent for the network's users, e.g. by migrating nodes or links in the physical substrate.

2. what are VNs used for?

VNs may be used to provide a dedicated network for its users which operates isolated from any other network traffic. They may be used to bring out completely new network architectures and protocols. They may be used to optimize resource utilization either via aggregating or via distributing resources used for a virtual network.

3. what are they key challenges?

for each challenge:
    - define the challenge
    - explain why it is hard
    - provide some references to those working on solutions

- Instantiation of virtual nodes and virtual links.  How can we a
 priori determine whether a virtual node exists or is actively running?
- Inter-domain-wide operation across multiple administrative domains.
 Comparable to the issues in QoS research we may be able to easily
 provide dedicated services to users within our domain, but providing
 these services inter-domain-wide may have some technical or business
 constraints.
- Dynamic and on-demand instantiation of virtual networks.  Instead of
 using a manual or semi-manual (e.g. script-based) configuration of
 virtual networks, it may be desirable to use signaling protocols for a
 dynamic and on-demand management of virtual networks.
- Network topologies. Talking about virtual networks, we may wish to
 not only control some virtual nodes and virtual links, but also create
 virtual network topologies, such as full-meshed networks, rings,
 trees, etc.
- Attachment of end-users---are they part of the virtual network? How
 do we cope with the mobility of end-users, again inter-domain-wide?
- Aggregation---to what level can we aggregate traffic by still
 preserving isolation and by not losing flexibility to migrate virtual
 nodes and virtual links to another location?
- Security---how can security be achieved? Is it an integrated part of the virtual network architecture and transparent to its users?
- AAA---using virtual network resources breaks down to actually using
 physical resources of one or more network operators.  How do we
 provide authentication, authorization, and accountability?  How do we
 provide these AAA services inter-domain-wide?
- Are Quality-of-Service guarantees an inherent requirement for virtual
 networks? If yes, to what extent?
- Providing the users the opportunity to create completely new network
 architectures and protocols inside their virtual networks---how do we
 cope with interoperability?  Or do we run the risk of creating an
 »inverse Internet«?


That's it for now.

Martin
_______________________________________________
vnrg mailing list
vnrg at irtf.org
https://www.irtf.org/mailman/listinfo/vnrg

--
Didier Colle
Ghent University - IMEC - IBBT
Department of Information Technology (INTEC)
Gaston Crommenlaan 8 bus 201, B-9050 Gent (Ledeberg)
Email: didier.colle at intec.UGent.be
MSN: didiercolle at hotmail.com
Skype: didiercolle
Tel. +32 9 331 4970
Fax. +32 9 331 4899
Mobile: +32 473 295655
WWW: www.ibcn.intec.UGent.be


Note Well: Messages sent to this mailing list are the opinions of the senders and do not imply endorsement by the IETF.