How we did it
 
|
|||
 | |||
|
Sixty days before the scheduled start of testing we invited 10 companies (V-SPAN, InView, WireOne, Sprint, Electronic Data Systems, Eyeball Networks, FocusFocus, First Virtual Corp., Reality Fusion and VideoNext) to participate. V-SPAN and EDS declined. VideoNext never responded to our invitation. In October, we invited NuVision, which declined.
For our guaranteed network service testing, we installed two identical H.323-compliant Polycom ViewStation FXs (running 6.0+ release of software) and 24-inch monitors at three locations (Placerville, Calif.; Southborough, Mass.; and Cambridge, Mass.). We also installed Tandberg 1000 videoconferencing appliances at two of these locations to test the interoperability of endpoints.
To test the Internet-based services, we installed Intel PC Camera Pros on Windows 98 SE, 2000, NT, Millennium Edition and XP operating systems running on 733- and 900-MHz PCs. For last-mile access, WireOne and InView included full and fractional T-1 circuits, respectively, provisioned by local exchange carriers (Verizon in Massachusetts and Pacific Bell in California). Sprint supplied T-1 circuits, but because of ordering and provisioning errors by Sprint, the services were never configured or tested as planned. The four Internet-based services entirely relied on the existing Internet access services at each facility: symmetrical DSL at one location and shared corporate network access from behind a firewall at the other two.
Testing phase
From Sept. 14 to Dec. 2, we tested user access via Web-based interfaces and videoconferencing application. We began with a service provider-hosted orientation, during which we were shown the service's features. We then used the videoconferencing appliances, applications and services in point-to-point and multipoint configurations. Our tests included scheduled multipoint calls, ad hoc calls, call forwarding (only available in the WireOne service offering), video operator assistance (only available in the WireOne service), gateway calls (on and off network, offered by WireOne and InView), data collaboration with a videoconference, and recording and streaming of a videoconference (using WireOne). We ranked video and audio quality on dedicated and converged networks and public VPNs.
The guaranteed network service providers were issued the same request for proposal, and their responses were evaluated on completeness, timeliness and professionalism.
We recorded and compared segments of video on the InView and WireOne networks at 768K, 512K, 384K, 256K and 128K using VHS Technologies in Cambridge.
With only three locations, we did not test scalability. And we didn't test privacy and security features beyond password-level protection.
Special thanks to Polycom, Intel, Tandberg, WildPackets Networks and NetIQ for their generous donation of time and equipment. The testing team included Jason Meserve, multimedia editor at Network World; Michael Pendleton of Cap Gemini Ernst and Young; and Christine Perey of Perey Research & Consulting.
RELATED LINKS
