The second component in Visual UpTime is the PAM (Performance Archive Manager), a Windows NT application that takes information from the ASEs, stores it in a SQL Server database and interprets it. Finally, the user interface for the system resides in the PAC (Platform Applicable Client). Both PAC and PAM could be integrated into the same server, though this is typically not the case. For very small environments Visual offers the MIC (Management Integrated Console), which combines PAM and PAC functionality into a single application.

There is a great deal of flexibility built into the Visual architecture. While only a single PAM is associated with a group of probes, the PAM database can be partitioned into multiple domains, and each domain may have its own access-control list. Thus a single PAM may provide multiple clients with data restricted to their own areas of interest and control. Planning and Reporting functions are available via a Web interface, so a service level manager can easily get reports with a standard browser.

Even more impressive is Visual's Burst Advisor, which takes the bold course of recommending changes to the CIR. On a per-PVC basis, Burst Advisor offers advice based on conservative, moderate and aggressive traffic management philosophies. In our tests, it correctly identified circuits that were overwhelmed with traffic and suggested increasing bandwidth (although it doesn't tell you what your new CIR should be). This is the only product tested that even attempted this high-level analysis.

The PAC application is broken into six major toolsets: Network Configuration, Performance Monitoring, Event Processing, Troubleshooting, Traffic Capture, and Planning and Reporting. The applications are well-organized and require little operator training to use. In the typical installation, operators will continually monitor the Event Configuration screen, watching for alarm thresholds. Visual does not provide an overall topology map of the network, focusing instead on management on a per-PVC basis.

UpTime caught almost every trap that we expected and required the least amount of preconfiguration work to manage the network the way we wanted. Unfortunately, the 4.1 version of the software could not trap based on excessive delay, nor could it periodically initiate a delay measurement. Instead, delay must be examined manually by an operator. This is expected to be addressed in the 4.2 version, which was released just after we completed testing.

UpTime is the only system we tested that was able to detect the latency of actual user data, based on correlating frames over the network. It doesn't use ping or otherwise inject its own traffic into the network to take this measurement. This approach is found in no competing system.

Also released in the 4.2 version is the ability to provide data to SNMP reporting systems, which has long been a point of criticism by Visual's detractors. While UpTime is still not completely open, it offers some data from the MIB II, Frame Relay DTE, DS1 and DDS MIBs. In addition, 4.2 adds the ability to configure probes via telnet, a feature still unavailable in NetScout probes.

While it doesn't have detailed application-monitoring capabilities, UpTime is not completely devoid of tools to analyze user data. The Traffic Capture application provides seven-layer protocol filtering and capture. The traffic-capture facilities look at the first 400 bytes of each packet, which provides enough visibility into packet headers for most network troubleshooting. Within the IP family, it is possible to set filters based on TCP, UDP, EGP, ICMP, IGMP, OSPF or any port number you wish to define. Unlike NetScout, the Visual probes can't generate traffic.

Visual CSU/DSUs cannot deliver management traffic over an existing user data PVC, unlike all the other products we tested (using Paradyne FrameSaver SLV in the NetScout case). Low-speed CSU/DSUs can monitor up to eight PVCs, in contrast to Paradyne's three-PVC maximum. Central-site probes can manage up to 108 PVCs. T1/FT1 CSU/DSUs may optionally feature multiple V.35 and DSX-1 drop-and-insert ports. Maintenance of the probes and their configurations is trivial, and probe firmware can be updated across the entire network with a single keystroke.

In the end, you won't need a huge time investment to get a payback from the Visual system. It intelligently presents far more information than what SNMP managers alone can offer, but less than a carefully configured and managed system using NetScout Manager Plus.

Adtran IQ View v. 1.0 (beta) with TSU IQ Probes
IQ View is a general-purpose SNMP management platform, with most of the features you would expect from an SNMP manager except a high price. Features include autodiscovery, topological mapping, SNMP polling, an alarm console, a MIB compiler and multilevel security. As such, it's not designed as a frame relay management application like Visual UpTime, nor does it offer the power of RMON and RMON2 that NSM+ offers.