For these tests, the TSU IQ CSU/DSUs were configured for in-band DLCI reporting. This approach can take a number of forms. First, Local PVC Management creates a "private" DLCI between a local router and the CSU/DSU, not appearing over the frame relay WAN. This option is also available with the Sync Research devices.

In the second in-band approach, Shared PVC Management sends management traffic over the WAN on an existing DLCI, stripping out packets destined for TSU IQs invisibly to the router/FRAD (frame relay access device) it serves. We used this mode for all our tests, and the Adtran devices performed flawlessly. In contrast, the Sync Research CSU/DSUs could not process moderately high traffic volumes (1 Mbps of IP traffic) using Shared PVC Management, or Piggyback DLCI mode in Sync lingo. All other systems passed this test.

Finally, a dedicated PVC can be set up over the carrier network that will transport management traffic. While this means higher ongoing WAN costs, it has applications for service providers and third-party management services that don't want to mix customer traffic with management data.

Unlike the NetScout probes, the TSU IQ devices support configuration via telnet. The TSU IQs support an optional Ethernet interface, although telnet and SNMP access over in-band DLCIs can save you the extra expense of installing Ethernet.

IQ View does not have the traffic-analysis capabilities of UpTime or NSM+ and doesn't see anything higher than the data-link layer. Thus, we weren't able to analyze our IP traffic loads or capture and decode packets. But IQ View and Adtran CSU/DSUs give you a lot of management horsepower for little expense--IQ View runs only $2,500, and the CSU/DSUs are price-competitive with nonmanageable products.

The version we received for review was still undergoing beta testing and did not come preconfigured with the default alarm configurations that are planned in the ultimate release. As a result, configuring thresholds was time-consuming though not difficult. In addition, we received many alarms that contained either no text or unreadable descriptions.

For reporting, IQ View comes with a simple trend-analysis package that generates both text and two-dimensional chart output. Unfortunately, the canned reports for service level verification were not ready in our beta release, although we were able to easily create our own reports to adequately track the most important service level metrics used in frame relay--utilization, availability and delay. Although the system is not built to track protocol distribution, top talkers and other important details of the load you offer to the WAN, when this product comes out of beta it should be worthy of consideration for cost-conscious managers.

Sync Research Envisage ControlSight and InSight v.1.000 with T-FRAP
We were excited after examining the specifications for the Sync Research frame relay products, but we were disappointed when we saw them in action. In our tests, the T-FRAP v. 2.55 probes we used could not properly process traffic and manage the network simultaneously. Under moderate loads we would lose the ability to manage the CSU/DSU-based probes, and under 1 MB of load we lost frame relay data communications altogether.

There is a lot of interesting technology in the Sync Research products. The VNIP (Virtual Network Interactive Protocol) offers the ability to map DLCI values to the appropriate PVCs automatically, a capability found in none of the other products. VNIP can also issue SNMP traps based on delay thresholds, a capability not found in our review version of UpTime. The Envisage InSight application provides most of the information you'd need to verify a service level agreement to a Java-based console.

Like Adtran IQ View, the Envisage NMS can serve as a general-purpose Windows95/NT-based SNMP manager, and it provides device-specific GUI configurators for Sync's products. A Unix version, TyView Plus, is available. Also like Adtran's products, management data can ride in-band over an existing data PVC, a "private" DLCI between router/FRAD and CSU/DSU or a dedicated PVC for management traffic. In contrast, Visual Networks probes can only communicate management data via an Ethernet interface, SLIP or a dedicated management PVC.

It was the PVC piggyback approach that we most wanted to see, because this approach reduces the ongoing cost to manage the network. But the T-FRAP could not keep the data network traffic running under a moderately heavy load. Immediately after turning up a megabit of IP traffic on our Wandel & Goltermann DA-30C Internetwork Analyzer, we lost management communications with the remote device, a problem that is acknowledged by Sync Research. The situation went from bad to worse after a few minutes, when all frame relay data traffic ceased to move over the network.

David Willis can be reached at dwillis@nwc.com.