Re:To Agent, or Not To Agent, That is the Question (Score 1)

Microsoft is actually making a decent stab at the problem with WMI, a sort of big brother to SNMP. Unfortunately the implementation is complex, non-standard, and up until now nobody has really used it for the type of remote instrumentation that this article talks about.

Actually, they WERE making a good stab at this - five or so years ago. Since then, the nature of where they're trying to go with this has changed, nearly the entire project-team has disbanded (and was reformed with a different focus), and they've stopped pushing for WMI-related WHQL requirements - thereby nipping the WMI-centric management idea in the bud.

To correct a few other misconceptions you stated... It really isn't all that complex; it's essentially an object-oriented database, with object instantiation, linkage, inheritance, an event system, and various other spiffy things. It IS most definitely standard; see the DMTF and their specifications for CIM and WBEM. And yes, SOME companies actually do make use of WMI for remote "agent" type system-monitoring; the company I work for makes extensive use of WMI for monitoring of remote servers, and proactively initiating Alerts that are sent out to our own servers letting us know when critical components of our customers' fault-tolerant servers have gone awry.

Don't get me wrong, Microsoft's WMI has had a LOT of problems in the past - the service crashing, memory leaks, lost notifications, etc. - but it's relatively stable now. And it's a fricking great idea in theory, as per the CIM/WBEM standards...