A Practical Navigator for the Internet Economy


We offer an in depth look at Ipsilon's IP Switch, an intriguing effort to use an innovative software approach to turn ATM switches into devices that can process IP efficiently. We have examined Ipsilon because of its likely impact on the outcome of the ATM versus IP debate. We also wanted to ascertain whether it promises any immediate dividend in the area of increases in routing power in the major Internet backbones.

We have also identified new backbone router projects at Netstar and Cisco. BBN is working on a DARPA funded monster machine and a new startup, Juniper Networks, is standing in the wings. It is likely to be nine to 18 months before any of these come on line.

In the meantime Ipsilon has developed a combination of hardware and software that allows it to throw out the cumbersome ATM forum software and plug a device it calls an IP Controller into almost any ATM switch. The software in the Controller is proprietary. However, it uses two Ipsilon developed protocols that Ipsilon has submitted to the IETF standards track process. These protocols enable the device to do connection oriented switching on long duration IP flows and routing on short duration queries such as DNS look ups.

The result, at modest cost, is a device that can handle upwards of several million packets per second. But, because of hardware limitations, the current design is appropriate for a "campus" WAN and not for a major Internet backbone.

What Ipsilon has done, however, is extremely important. For it holds out the promise breaking the ATM Forum standards log jam and of making hundreds of millions of dollars worth of ATM switches useful within the IP world. While Ipsilon offers its own software-hardware combination, it is also aggressively and successfully licensing its software to most major switch manufacturers.

We publish interviews with Larry Blair the VP of Marketing for Ipsilon, Noemi Berry of NASA Ames, who has been testing the Ipsilon switch for 6 months, and Peter Newman one of the Ipsilon architects.


Peering and transit among the majors is seldom discussed in public. When a lengthy debate broke out on NANOG early in May we paid careful attention. We present an edited transcript that shows some of the difficulties that smaller ISPs face in arriving at NAPs and trying to negotiate peering with the top "six".


Bob Moskowitz describes continuing efforts to place auto parts suppliers in a future Automotive Network Exchange composed of certified ISPs. Northern Telecom has announced it is using Java to develop an Internet client server telephone. Northern Telecom, AT&T's Lucent technologies and two smaller companies are thought to be developing gateways that will take Internet phones into the PSTN.


O'Reilly has published a whole family of books dealing with web based technology. The best of these is Designing for the Web - a general introduction to the use of HTML. We offer brief looks at this and four other volumes that together make a definitive library for those involved in mounting or maintaining web infrastructure.


An industry insider offers an assessment of the forces at work in each company that will either make or break the merger.


Last fall we promised readers a fresh look at MercerNet, the NTIA award that will bring interactive TV and Internet to Mercer County N.J. high schools and libraries courtesy of Comcast and $700,000 of federal money. We detail nine months of effort to get the project's Principal Investigators to talk with us.

The bottom line is that, in the absence of strict criteria from NTIA, forcing future grant applicants to show that they have an open process designed to include the entire community that they purport to serve, we have concluded that the practical effect of these grants is to embed large corporations in positions of control over a community's information highway infrastructure. We trace a process in our own back yard where two wealthy school districts and the local community college have allied with Comcast Cable to build a county-wide video and Internet system that is given, initially cost-free, to the rest of the county.

The problem is that the grant application process takes on faith the intentions of the Principal Investigators to run an open and democratic process involving, not just the school and library bureaucracy, but all the taxpayers of the county. In the case of MercerNet, the PIs explained to us that since they all had full time jobs involving things other than MercerNet, they had no time to become involved in a bottom up democratic process of building the network. Instead they have worked by committee within the school systems they represent and are now almost half way through a grant period where the general public, that will have to pay the on-going costs of this new technology forever, has had no input into the system they have created.

Indeed it is our concern that the federal government has aided and abetted a process whereby a private corporation, beholden to its stock holders and not to the residents of Mercer County, has been entrenched as the principal source of technology expertise within school systems where administrators have no means of making independent technology assessments. Having volunteered our time to our own school district technology planning committee as a part of our research for this story, we have direct evidence where of we speak. The blind are leading the blind and the federal government, courtesy of the efforts of local technocrats, has inserted a large corporate fox into the local school and library hen house.

Thanks to the refusal of the PIs to take their implementation process public, the residents of Mercer County have been able to build no independent means of establishing their own future technology policy. NTIA should add community involvement criteria to its latest round of grants. Applicants should have to demonstrate commitments from local businesses and computer geeks (who are not a paid part of the project) that they'll contribute "sweat equity" to the local efforts. Projects that do not enjoy broad support from local business, government and education should not in the future be funded. If MercerNet had had to meet such criteria it would have been a very different project.

We contend that the Clinton Gore NTIA grants have undermined the ability of Mercer County to ever build a technology infrastructure that it can own and control and use to its resident's benefit rather than that of a huge and distant corporation. The process is broken and needs to be fixed or abolished.

The full text of this article is also available as MercerNet.


AGIS is being sued by Lynn Bateman, a former share holder of Net99.