Network Computingwww.networkcomputing.com

This product's management interface was just about as easy to navigate as ZyXEL's offering. Although there are more screens to traverse to get the job done, the layout is fairly easy to use, and configuration is painless. Like the Prestige 128U Extra, the PACESETTER SOHO handles packet filters and adjusts Multilink PPP parameters with ease. We fully monitored the router functions and logged various levels of activity using a simple telnet application.

Jeff Newman can be reached at jnewman@nwc.com.

How We Tested

In our San Mateo, Calif., lab, we created a separate LAN segment to represent our remote office. On that segment, we put Windows NT 4.0 Workstations running on a Dell Dimension P90, using Ganymede Software's Chariot benchmark end point. Each router was placed on the LAN to connect the segment to our central office via BRI, which was hosted by a Cisco AS5300 Access Switch via PRI circuit. At the central site, a Windows NT 4.0 Server running on a Compaq ProLiant facilitated and monitored data transfers and served as an end point for the Chariot benchmarks.

For each test, Multilink PPP calls were initiated to the Cisco AS5300 with Stac LZS compression enabled. Bandwidth on demand was then tested by executing a series of file transfers that elevated the traffic above the preset traffic threshold of 50 percent utilization of the BRI line for 10 seconds. Once the channel coupling was established, a series of Chariot benchmarks were initiated for FTP transfers from the NT Server at the central site to the NT Workstation at the remote site.

The series of FTP transfers consisted of highly compressible data and precompressed data, which indicated how each router would react while trying to decompress the data that was already compressed in a different format (for example, it might back off to save performance or try to force the decompression and fail). Each benchmark series was then executed in reverse (transfers from the remote site to the central site) to measure each router's processing power and behavior when presented with the task of compressing data that was already compressed.

To test functionality, we connected Conair speakerphones to each router that supported POTS. Once connected, calls were placed to and from the routers to normal telephone extensions served by Pacific Bell. Incoming call routing, bumping and blocking was tested by placing calls from our lab to the ISDN routers during different data channel utilization stages.