It turned out that the tech support people were right in one sense. The trend lines were drawn directly over the actual usage lines, providing an effect similar to the validation value HP's PerfView Planner supplied. The only difference was that we couldn't vary the amount of data projected as compared to the amount of actual collected data, as HP's PerfView Planner lets you do. This left us to accept or reject the projection with no way to verify its accuracy.

Unicenter TNG includes a charge-back application, which we were eager to try. This Excel macro is designed to create invoices. We got as far as dividing users into groups but we couldn't create invoices. Tech support had no explanation, and as of press time the problem was unresolved.

During our struggles with the chargeback application, we spent a fair amount of time trying to get all our agents to supply data. Unfortunately, we learned that the macro only supports Unix agents (a fact not mentioned in the documentation).

PLATINUM technology ProVision Network Monitor and Server Vision
PLATINUM's ProVision covers a lot of ground. The ProVision suite contains familiar enterprise systems management applications, including scheduling, database management, server management and software delivery. It also includes common messaging and repository services. Server Vision is an excellent systems performance diagnostics tool that combines good canned functionality with excellent flexibility. And Network Monitor--an event management engine for SNMP alerts--proved to be an excellent tool.

Like CA and Tivoli, PLATINUM uses a common framework to provide a total management solution. PLATINIUM'S products began as a collection of best-of-breed point products. Only recently has the company begun selling this collection of products glued together on top of a common platform.

As with CA's Unicenter TNG, the range of these products exceeded the scope of our review. Also like Unicenter TNG, ProVision's price tag is on the high side, and the individual products, while strong, have occasional lapses that make it practically a necessity to buy the entire suite. For example, on its own, PLATINUM's ServerVision is a strong tool for diagnostics and historical analysis of systems performance statistics, but it lacks a NetWare agent. According to PLATINUM, Network Monitor is such a strong event-correlation engine and so many canned events are based on Novell's ManageWise agent that it makes sense for customers to pay the additional cost for a more complete package.

The diagnostics tools offer a nice blend of real-time and historical statistical views. The agents are installed with default statistical metrics, which are immediately available for viewing by collections--CPU, file I/O, interrupt, memory, disk, processes or network statistics.

We chose an agent and expanded an Explorer-like interface that offered many collections of data. With a right-click selection of a chart, we were able to drill into any historical statistics. The ability to choose any of these general data groupings is enhanced by related extended statistics, which detailed the chosen group along with related information that might have a bearing on the original performance group.

It was in this area of related statistics that PLATINUM's maturity becomes apparent. It shows in the way statistics are displayed and in its excellent online help module, called The Advisor. This contextual help not only commented on what the statistics were saying, but also gave probable causes and diagnostic prescriptions for resolving problems.

During our testing, we had the opportunity to upgrade to the latest version, which offers tighter integration between the configuration and diagnostic viewing portions of Server View. In this upgrade, you can right-click on an agent and there's a new launch point into the Historical Configuration Editor (HCE).

The HCE is used to modify PLATINUM's long list of scans. We found it easy to create our own scans to monitor a textual parse of a log file. The HCE has a predefined set of templates that users can modify.

We were disappointed that the common services couldn't intelligently handle failed communications with downed agents. When an unavailable agent was selected, the console would just hang, waiting for the agent to come back up. It should have immediately timed out and returned an unavailable status.