IP: 2 notes -- one by me -- Internet II - a set of concerns --

[Dave Farber <farber () central cis upenn edu>]

Internet II - a set of concerns


Professor David Farber 


The Alfred Fitler Moore Professor of Telecommunication Systems 




University of Pennsylvania




Oct 8 1996












This note is a follow-on to a document titles INTERNET II dated July 20,
1996. In that note I questioned some of the assumptions that were behind
the plans to build a new NREN for the University and College environment.
This note sharpens these concerns in light of the current proposals. 


I believe we need to focus at the next stage in developing the US
infrastructure and base it on the emerging new technologies for
distribution rather than just speeding up today's internet. I also believe
that we should stop one dimensional thinking of wired IP land nets and
attack with vigor the newly arriving wireless, digital satellite along with
things like cable and XDSL technology. >From the corporate and university
perspective, I believe it is time to focus first on complete campus
communication solutions which will include the merging of the telephone,
computer and video structures into one common system and will utilize
wireless personal communication modules (modular multimedia PDAs) to allow
the community to more fully utilize the capabilities of the infrastructure.
Classes, video and supporting data components, will be sent to the extended
campus community (which may include other Universities, colleges, industry
and people's homes) via XDSL and Cable Modems in the local environment and
DBS channels in the wider and international arena.  


I believe that the following statements are true and should form the
guiding principles on which the INTERNET II is based. They are: 
*  the main reason for the poor performance of the current internet for
researchers is usually due to congestion in the core of the net not the
campus. Creating another separate network for the R1 and R2 schools will
for a while alleviate that problem (at a high cost) but in the short long
run, non-academic traffic will route over the new network since history has
shown that academia and the rest of the nation are not separable. My belief
is that the bottlenecks in the core of the network will cause problems for
the new technology access providers such as cable or xdsl and slow down
their introduction. Creating an immediate reason to build up the core
internet will in the long run payoff with a faster adaptation of the new
technology with it's attendant benefits to the nation. 




* the University community can not exist in isolation. It is an error to
create a unique isolated network for it's use. That is not to say that it
does not need mechanisms to insure bandwidth-latency requirements for
research can be achieved. However similar mechanisms are needed in
industrial research environments as well as future consumer services. 




* the INTERNET II should not attempt to bypass commercial internet service
suppliers but should create a marketplace for advanced network services to
be offered by such organizations. Only at the bleeding edge of research
should the government support the provisioning of technological R&D and the
establishment of test beds. OC12 backbones are not bleeding edge. Managing
to deliver OC12 to workstations with quality of service capability is near
the edge but required a major investment (or re-investment) in campus
infrastructure and a major re-investment in student and faculty
workstations if it is not to be a toy of the researcher. Whether the
Federal government is willing to foot the bill for such a re-investment is
highly questionable and whether the University by itself can and will is
also questionable. .  




* service criteria including availability, quality of service (QOS),
security etc. should be a basic part of the engineering of INTERNET II. Any
RFPs and resultant contracts should incorporate such criteria and demand
that they be passed onto subcontracts. The aim is to ensure end to end
performance given an environment where no single contractor will be
responsible for everything. The research community should be tasked to
develop criteria, possible technical ways of insuring performance as well
as methodology for the measurement and reporting during operation. 




* an operational framework should be planned and implemented to oversee the
evolution and performance of the INTERNET II. I would strongly suggest that
we ask leaders in from industry who have been there and understand such
issues to help plan this framework. I strongly suggest people of the
caliber of Mort Myerson ex of Pirot Systems and EDS and Colin Crook CTO of
Citibank be asked to organize such a group. Both Mort and Colin have "been
there" and know how to organize and manager large communication systems and
how to deal with suppliers and operators. 




* the Universities would form a critical force in NI2000. They would be the
intellectual focus of the NI2000 in their communities and would be the
technology transfer vehicle for the deployment of INTERNET II in the
communities they service. In many cases this might entail the Universities
being the hosting center of the communications systems allowing the
community to have the use of a part of the external bandwidth while in
other situations a community hub will become the connection point for all
the aspects of the community with quarantined segments of bandwidth and
services sold to all comers. In this case the University would be the
intellectual focus of new and interesting applications as well as a
supplier of educational services for part of the community. 




* universities should attempt to construct a sound business and economic
model for Internet II. Although some subsidy may be necessary to cover
setup costs, ultimately the technology must pay for itself if the
university experience is to be extensible to the private sector. 




*  the role of the federal government should be to help establish the rules
as outlined above and to provide lubrication by the funding of Model Cyber
Communities (MCC2000) where all aspects of INTERNET II are deployed in a
community with the support of all aspects of the community life. The
MCC2000 would provide the opportunity for the business world to support
INTERNET II by taking part in the planning for and provisioning of
communities and the industry's willingness to take part would be a basis on
which awards would be made by the Federal Government to support the MCC2000. 


Posted-Date: Wed, 16 Oct 1996 07:22:47 -0400
Reply-To: rgm3 () chrysler com
X-Sender: rgm3 () pop3hub is chrysler com
Date: Wed, 16 Oct 1996 07:18:35 -0400
To: farber () central cis upenn edu
From: Robert Moskowitz <rgm3 () chrysler com>
Subject: Re: Internet II - a set of concerns


At 06:02 AM 10/16/96 -0400, you wrote:


Check out http://www.aiag.org/anx


> From the corporate and university
> perspective, I believe it is time to focus first on complete campus
> communication solutions which will include the merging of the telephone,
> computer and video structures into one common system and will utilize
> wireless personal communication modules (modular multimedia PDAs) to allow
> the community to more fully utilize the capabilities of the infrastructure.


This needs to be parrallel.  Senior management is already at the extended
enterprise, but our networks are not.


> *  the main reason for the poor performance of the current internet for
> researchers is usually due to congestion in the core of the net not the
> campus. Creating another separate network for the R1 and R2 schools will
> for a while alleviate that problem (at a high cost) but in the short long
> run, non-academic traffic will route over the new network since history has
> shown that academia and the rest of the nation are not separable. 


The AIAG recognized this over a year ago and has opted for a path the work
'within the system'.


> * the University community can not exist in isolation. It is an error to
> create a unique isolated network for it's use. That is not to say that it
> does not need mechanisms to insure bandwidth-latency requirements for
> research can be achieved. However similar mechanisms are needed in
> industrial research environments as well as future consumer services. 


Where does the auto industry end?  Labor dept says that 1 in 6 americans
work auto related.  So even though we are starting with our productive
parts suppliers (~8,000 companies in our supplier databases), we have no
intention on limiting the scope of the ANX.


> * the INTERNET II should not attempt to bypass commercial internet service
> suppliers but should create a marketplace for advanced network services to
> be offered by such organizations. 


We require competition by the best of the commercial providers, but we will
certify and monitor them.  Not many should be able to 'make the grade'.
Requirements need to be demanding.


> * service criteria including availability, quality of service (QOS),
> security etc. should be a basic part of the engineering of INTERNET II. Any
> RFPs and resultant contracts should incorporate such criteria and demand
> that they be passed onto subcontracts. The aim is to ensure end to end
> performance given an environment where no single contractor will be
> responsible for everything. The research community should be tasked to
> develop criteria, possible technical ways of insuring performance as well
> as methodology for the measurement and reporting during operation. 


This is our approach.  We are negotiating right now for our 'overseer'.


> * an operational framework should be planned and implemented to oversee the
> evolution and performance of the INTERNET II. 


We are doing this for our effort.  Hopefully in a month we will have a
contract.


> * the Universities would form a critical force in Internet II. 


The OEMs are taking a similar position, even though it is the 1st and 2nd
tier suppliers that gain the greatest benefits during the initial phases.
The big 3 are funding the start up costs for the overseer, and will most
likely have other expenses.


> * universities should attempt to construct a sound business and economic
> model for Internet II. 


We are likewise struggling with this.  Our VPs and suppliers are demanding
it.  We expect to have a good handle on it after our pilot in around 6 months.




Robert Moskowitz
Chrysler Corporation
(810) 758-8212