One-thousand-one. One-thousand-two. …
Mainstream commentary on application performance management (APM) frustrates me often. Elaborations of “business alignment” or ambiguously-phrased surveys about categories of influence are too abstract for me–and they too often neglect the single dominant, almost “physical” fact that anchors consequential APM practice:
The conceptual basis of APM is important, of course, especially as we enter the “third generation” of APM tools. “Application Monitor” will return in 2013 to consider this “reinvention of the category” in more detail. For me, though, all philosophizing about strategy is secondary to the central realities of end-user experience (EUE):
These realities lead to a couple of conclusions: first, the temporal scale of application delivery is a fundamental that everyone in information technology (IT) needs to feel at all times. Successful mass-market applications might be fit in time the way dogs or horses fill up space, but they’re not elephants or mice. The few seconds we have to deliver an application experience set the gage for everything computers can or should do.
The second conclusion is that acceptable performance is delicate: the mathematics of complex interactions teach that small, routine variations in performance of individual connection elements like load balancers or loaded networks can result in catastrophic degradation of EUE. In such a situation, organizations must monitor results closely with tools that diagnose root causes in real-time. Whatever its name or budget entry, APM is more necessity than choice.
Your IT systems need to show results before distractable end users wander away. Anything else turns all the people, hardware, and plans of IT into a sterile expense.