The central office site consi sted of a Compaq Computer Corp. ProLiant 4500 running Windows NT Server 4.0. An Ascend MAX 4000 Primary Rate Interface (PRI) router running version 5.0Ap13 firmware hosted its connection. Connections were established in a variety of combinations, including Password Authentication Protocol (PAP) and Challenge Handshake Authentication Protocol (CHAP) authentication, compression negotiation, dial-on-demand, and Multilink Point-to-Point Protocol (PPP).

We used Ganymede Software's Chariot benchmarking software (hosted on another Windows NT 4.0 server at the central office side of the connection) to test both compressed and uncompressed IP routing performance between the Compaq ProLiant 4500 and our remote Dell Dimensions. Tests were run for both compressed (.ZIP files) and uncompressed data. The same series of tests were executed in reverse to test the processing power and ability to route, compress and decompress in both directions.

In addition to providing Ethernet ports, ADC Kentrox, Cisco Systems and Ramp Networks provided us w ith units that also offer plain old telephone service (POTS) interfaces. To test the reliability of these interfaces, including dial-in/dial-out and ring operation at variable phone line distances, we used two ConairPhone (Model HAC CM200) POTS-powered speakerphones. All three units were able to power both phones, off a single port, at distances of 50 feet over standard phone lines and to drive a single phone at 100 feet. Additionally, we tested these units' ability to power an older model Bell (Model R82NA11) mechanical ring Touch-Tone phone. The ADC Kentrox unit was able to ring it at 100 feet, while the Cisco 776 and Ramp Networks WebRamp wouldn't ring it at as little as six inches. In addition, the Cisco 776 couldn't complete a simple dial-out at any length of wire.