The mainframe, no doubt, was an easier nut to crack. First, the hardware was instrumented for management. Not only could you read status information, you also could run diagnostics and actually put the status information you'd collected to good use. IBM and others were accountable for a soup-to-nuts solution, so it was in their best interests to make it possible to detect and correct problems with their big iron systems.

Network vendors weren't of a mind to provide that kind of management. Their customers weren't looking for it, and if they built it in, they probably would have priced themselves out of the market. Systems vendors never have been keen on letting an outside management package run the show--and even if they had been, the decoupling of operating system from hardware changes the task immeasurably. Of course, some vendors, like Cisco, did a good job instrumenting their hardware, but then didn't provide mechanisms to do much with that information, often choosing to keep the interface to useful systems management private.

SNMP was an extremely poor attempt to create a framework for management. Its standardized MIBs didn't provide meaningful statistics, security or the mechanism for managing the components of the network. Proxies that reside on systems often present data in less than useful ways. The growing pains endured by the industry have been excruciating. We now have a patchwork of standards that, if taken as a whole, implemented just right, and studied and coerced, can actually get you to a managed enterprise.

Getting it all right makes dancing on the head of a pin look like child's play. Most vendors have reverted to an agent strategy to deal with all the various types of software, hardware and devices on a network. Often, they have barely cracked the monitoring nut, let alone attempted genuine end-device management. This leaves you with weak systems and a weak network management strategy, or one that relies heavily on point management products to get the job done.

Running Scared or Hanging Tough? This would scare the hell out of me. IS managers are responsible for implementing a system that is the lifeblood of a modern enterprise. Corporate America now spends more on information systems than anything else, and the systems that connect everything can't be effectively managed as an asset.

We are tough on enterprise management products, perhaps tougher than we need to be. But what really amazes me is that after we review one of these products, we get almost no reader feedback indicating that we've missed the mark (though on occasion a few vendors have lit into us in a big way).

We've heard plenty from people who live with and love particular point management products. These products make a lot of sense. They have one purpose: to make a piece of hardware work right. This is a problem that vendors can get their arms around and a solution that you and I can easily appreciate. But who really has a product that answers the need to know about the IS infrastructure's health as a whole?

I want to hear from those of you who have implemented the major enterprise management packages (for systems management or network management) and love them. I want to know what they do for your company and what it took to get value out of them. And I'd like to hear what you would do to make these systems even better. And if you absolutely hate 'em and need to vent, go ahead and send your rants my way too.

Send your comments on this column to Art Wittmann at awittmann@nwc.com.