It’s a good time to invest in application performance management (APM) tooling. That’s how it looks to me, at least, and several items in the news reinforce that sentiment.
APM has responsibility for expression of daily operations reality in terms of business value. While successful organizations always made information technology (IT) choices based on long-term organizational goals, aligning those two was simpler in the slower-changing, more “siloed”, smaller-scale past. Now, not only are intuition and inertia insufficient guides to IT execution, but APM itself is a specialty best left to experts.
That’s at least part of the background I see behind the “2012 Application Performance Management Outlook Survey” from Quocirca. That survey, to my surprise, reported “that the overriding priority for IT managers was application performance.” While the statistical methodology is a bit delicate–as I read the numbers, it looks as though a slight adjustment in the categories pushes “cloud” to the top–APM certainly is getting a lot of attention.
Ovum Principal Analyst Michael Azoff attributes this “period of revitalization” to “waves of Agile development and … DevOps, as well as Cloud Computing, virtualisation, and mobile device adoption, and technical innovation within the APM industry.” Notice the body of his posting doesn’t even mention bigdata or compliance, two more powerful trends which themselves demand plenty of APM!
Challenge and opportunity
Lists of trends like this quickly overwhelm IT strategists; there’s only so much we can take into account at one time, which is itself of course one of the reasons for the importance of APM. What I find interesting about the highlighted trends is the degree to which they all have a multi-dimensional impact on APM. Consider virtualization as an example: certainly the availability of virtualization techniques changes the way we manage our datacenters, and heightens the importance of accurate performance metrics. At the same time, virtualization challenges existing models of APM tooling; timespan-based APM monitors must be re-tuned, or at least re-interpreted, when deployed in virtualized contexts. Notice also that many parts of appliances now turning up on virtual machines (VMs) are not coded in Java Enterprise Edition (JEE) or .NET, the traditional centers of APM capability.
That’s not all: a wider application lifecycle management (ALM) perspective recognizes that virtualization promotes the rapid cycling and responsiveness Agile and DevOps promise, and thus further amplifies the importance of measurement. Virtualization makes large-scale off-line testing practical, and also makes vertical and horizontal scaling more feasible, because virtual environments have good replicability.
Similar analyses apply to nearly all the other important trends affecting IT now: mobility and DevOps and the others not only change the numbers we need from APM, but they change the nature of APM itself.
This combinatorial complexity puts a premium on disciplined measurement, because so many variables are changing simultaneously. It pays to have APM which is “current”–aware of the latest practices in mobility, virtualization, and so on–and broadly capable, simply because APM results now affect so much of IT. Anything less amounts to running “in the dark”, because it’s simply impossible to keep track of all that affects performance from one end out to our customers.