The following text is copyright 1997 by Network World, permission is hearby given for reproduction, as long as attribution is given and this notice is included.

How important is fast?

The next generation Internet initiative announced by President Clinton during the election campaign last fall is turning out to be something quite real. A draft action plan is now available for public comment. ( http://www.hpcc.gov/ngi-concept-08Apr97/) Comments are due by May 15th.

This is an ambitious plan, supported by $100 Million or so of federal funding. It has 3 main goals: high-performance connectivity, advanced network service technologies, and to demonstrate revolutionary new applications.

The high-performance connectivity goal has two sub goals: first, connect up 100 or more sites through a network with interconnect speeds of 100 Mbps or more, and second, connect up a smaller number of sites at rates greater than 1Gbps. The first sub goal will be accomplished by strengthening the existing NSF very high-speed backbone network service (vBNS) along with NSF's high performance connections program and perhaps by working with the higher-education driven Internet II project. The second sub goal will involve building a new very high speed network.

The advanced network service technologies goal will develop and demonstrate those advanced network service technologies that will be required in future data networks to support next-generation applications. The areas of interest include quality of service (QoS), security, robustness, network management, routing protocols, computer operating systems, distributed applications environments and others.

Finally, the goal of demonstrating revolutionary new applications will explore a few targeted next-generation applications and will be a demonstration of the sorts of network-hungry applications that will be possible using the high performance connectivity and advanced network service technologies.

The NGI initiative follows in the footsteps of a long history of US federal government sponsored programs that have helped create the technology that has made it possible for the Internet to have grown to the scale and importance that it has to date. With its ambitious goals this initiative clearly has the potential to develop technology that will be critical to the continued growth in scale and uses of the Internet. But I do admit to a nagging worry.

It does not seem to me that enough emphasis is being put on goal number 2--developing new network services technologies to provide a more useful Internet. Clearly research in ultra high speed networks is quite important but it is not just pure network speed that we need to support. A very real need is to develop the ability for the network to handle variations in demand and to scale both up and down in speed. This leads me to wonder how much better the current Internet would now be if more research money had been spent on IP in the last few years.

It is not clear yet what the balance of funding will be for each of the NGI goals but it is clear that building a new network to provide gigabit and higher data rates is a very expensive proposition and it would be a shame if we were to dedicate a disproportionate share of the NGI funding to this effort at the expense of working on the network services technology that is badly needed for networks of all types and speeds.

disclaimer: Of course Harvard has great QoS and has demonstrated significant robustness so the above worry must be my own.