What are the limits to APM (application performance management) and application monitoring in general? In postings like yesterday’s “Sky’s the limit …“, the “Real User Monitoring Blog” consistently argues that we need to look at APM inclusively. Well-done APM accounts for networking, storage, and servers; it occupies a strategic position in the organization; it encompasses perspectives from the deeply technical to the utterly subjective; it understands not only J2EE and one particular database, but also .NET and several alternative data repositories; and so on.
That’s not all. Senior industry analyst and performance management specialist Michael Azoff makes the case that we’re still not looking broadly enough at APM. For him, security is overdue for APM attention. Especially at a time when so many applications integrate external application programming interfaces (APIs), cloud-based resources, and otherwise “run outside the firewall”, Azoff urges that “security should be something that developers think about right from the start.” With only a few vendors have begun to integrate security in APM, Azoff says his aim in regard to security-APM integration is “creating awareness.”
Another boundary that APM needs to negotiate is that of Big Data. In “15 Top Reasons Why APM Deployments Fail“, multiple authors mention the burden of the “huge”, “vast”, “enormous”, and “unmanageable” volume of data APM generates. The solution: leverage the capabilities Big Data approaches have built up “to automatically learn the environment’s behavior and detect deviations … before performance impacts systems and users …”
Cloud hosting clearly shapes both security considerations and Big Data, as the articles referenced above hint. Identity is another crucial dimension in the middle of transformation by cloud capabilities. For Jeffrey Kaplan, APM is the natural place to integrate identity management precisely because identity affects more than just security: it’s crucial for “… monitoring application usage levels which give corporate decision-makers important insight into which applications are used most often so they can make better purchase decisions.” The result: “… a key new value proposition for identity management …”
Suppose you accept all these recommendations: you agree that APM pays off best when it comprehends security, Big Data, identity, and perhaps even others. Does that mean that you generate a Request-for-Proposal (RFP) to potential vendors that demands all these parts?
Probably not. A few other sophisticated markets, including SDLC (software development lifecycle), have already accepted that successful products need to be frameworks, open enough to allow plugins or integrations. Azoff emphasizes that “It is possible to get a unified view by choosing point solutions that together provide that unified view … It is possible to have a set of point solutions that together cover the whole spectrum of performance issues.”
Focus and fitness are essential, of course. Scope your APM projects to be manageable in size. At the same time, recognize how much of a load your APM workhouse can eventually carry for you, once you give it the chance to stretch its legs.