FRADs Make Sound Sacrifices To Get The Data Through

By Jeff Newman   We were skeptical when we first heard of voice over frame relay. Frame relay was not designed to provide the quality of service guarantees that time-sensitive traffic requires; rather, it's a data service that works well for interconnecting LANs. It is one thing to hear the miracle of dial tone over frame relay at a trade show, but it is quite another to trust your voice traffic on a real network. Sure, you will save money, but is the current technology up to the task at hand? Several vendors insist it is. To see if they are right, we recently tested several voice and data FRADs (frame relay access devices) at MCI Developers' Lab facilities in Richardson, Texas. We tested Acclaim Communications' 3200V Access Device, ACT Networks' NetPerformer SDM-9300, Alcatel Data Networks' 1100 QIK Model 305, MICOM Communications Corp.'s Marathon 3K Integration Multiplexer and Motorola Information Systems Group's Vanguard 320. Our findings: Small offices using these devices will have to sacrifice some voice quality. To view the Report card. Nuera Networks and Memotec, two players in voice over frame relay, were regrettably absent from our testing. Nuera Networks does not offer a FRAD that routes TCP/IP. And though Memotec elected to participate and met our testing criteria, we were unable to resolve technical difficulties in time for our tests. Strong PBX Support All of the units we tested for this review offered strong PBX connection support, including support of FXO (foreign exchange office), a trunk loop start connection that emulates a single-line telephone to central office lines or PBX stations, FXS (foreign exchange station), a station loop start connection to a standard, single-line telephone instrument, or a loop start trunk circuit of a PBX, and 2-wire/4-wire ear and mouth (E&M), a connection to provide a tie line trunk circuit to connect between PBXes or other voice switching system. Because all of the FRADs tested support these standards, we eliminated the PBX as a variable in our test infrastructure and connected our analog telephone lines directly to the voice ports of each unit. Can You Hear What I Hear? Data performance is easy to test, but sound quality is largely a matter of perception. What's acceptable to John may be torture for Mary. To quantify the FRAD's voice quality, we used a standard technique employed by the telecommunications industry known as CMOS (comparative mean opinion score). This let us test the quality of each FRAD 's voice functionality compared with a public long-distance network voice sample, and the quality of each FRAD's voice compared with the other FRADs we tested (see "How We Tested Voice & Data FRADs" on page 86). For our tests, we assembled a random group of 32 men and women from Network Computing's San Mateo, Calif., office and had them evaluate the voice quality of each FRAD compared to the same transmission over the public telephone network. Voice samples were gathered from each unit while the devices simultaneously routed various-sized IP data streams. None of the FRADs were able match the quality of a standard long-distance network call. ACT Networks' NetPerformer SDM-9300 achieved the best voice quality with an overall CMOS average of -0.70. All the FRADs are capable of achieving better performance by tuning voice/ data handling variables. But higher throughputs generally mean sacrificing voice performance, or vice versa. The voice and data prioritization, segmentation and bandwidth settings we used were taken from the vendors' recommended settings.

The Voice and Data FRADs features chart , in Acrobat format. How We Tested Voice and Data FRADs, in Acrobat format. For the Side Bar on How We Tested Other Reviews Secure E-Mail Clients: Not Quite Ready for S/MIME Prime Time. Stay Tuned. By Dan Backman The Messaging Choice Is Clear By Dan Backman Related Links Building a Frame Relay Network Your Guide To Finding A No-Fret FRAD Frame Relay Makes Its Voice Heard Make Room For Frame Relay