NSF Wireless/Mobile Planning Group Workshop

Home | Agenda/Slides | Abstracts | Position Papers

WMPG Technical Proposal

Internet Architecture White Paper

Presentation Abstracts

Position Papers

Breakout Sessions

Wireless Implementation

WMPG Site VIsit slides draft for comment

Plenary Talk

A strategy for Continually reinventing the Internet
Larry Peterson


Questions: Need clarification about what you mean by architecture. We need architecture of back bone or end to end architecture ?

Larry: An end-to-end architecture

This talk does not have a wireless perspective; the red block contains everything you need to know about wireless. PlanetLab aims at virtualization of the backbone. Diff parties will control the 3 rd party software.

Dave: practically for this workshop – specific MREFC proposal. Need virtualized infrastructure; need core to reach out towards the edge. How to make this happen? challenge to L:

Reference wireless network.

Testing new hardware requires setting up the hardware on the rack.

How do I integrate new hardware with platlab or NLR.; virtualization model

Ans: virtualization happens w

Dave: the virtualization model may not serve someone who wants to observe bits on a pipe.


Optical switch overlay; connect to one of the bb routers in the rack directly; Gary Minden thought this was cheap virtualization; Different types of for virtualization in differ parts would be desired. An architecture which has packets at some levels and aggregation else where leads to a messy testbed.

Peter: Virtualization picture with red blocks; virtualized wireless networks is an important topic of discussion

Mario: PlanetLab has dedicated pipes for each experiment. It does not allow studying interaction between experiments; This is inconvenient unless you assume separate chunks in Internet in future.

Gary Minden: What is the final value of such testbeds to the country?

Larry: correct security risks.

Wade: how to ring in the common man?; Larry: plant lab already doing that.

Badri: Cellular industry – ossification taken to an extreme;

Ray: two other projects building such testbeds/

Ray: Wireless is represented as a tiny link in your slides. But one cannot draw a boundary of influence with wireless networks. Meaningful virtualization of radio devices also needed.

John Heidemann: treating wireless networks as a edge leads to loss of opportunity. Need a wireless domain for virtualization

Ray: Wireless Adhoc Multihop cannot be encompassed here. So cannot write wireless as a separate cloud.

Nitin Vaidya: Security: what can you not do with overlays ?

Larry: Cannot represent DoS.

Dick Grunwald: Level 3 that bridges VLANs with MPLS. Why do you need to deploy over NLR ?

Larry: Cost of doing with NLR much smaller than with MPLS.



Instructions for Workshop Participants

Directions to WINLAB Tech Center II facility

Directions to Workshop Dinner on Aug 2

For problems or questions regarding this web contact webmaster@winlab.rutgers.edu.