In most application teams, the approach to application monitoring is to watch everything in production environment – all the time – to find customer applications issues that have slipped through the cracks.
This produces reams of deep-dive metrics that are then used for post-mortem analysis by developer experts and middleware mavens, in much the same way doctors study patients to prescribe medicines and assist in recovery.
However, time and time again this approach has been proven ineffective, because the Application Performance Management (APM) solution used is simply too metrics-oriented and descriptive – not prescriptive or offering much in the way of problem resolution.
Needle and haystack problem
By fixating on just how much descriptive data they can get from logs, app servers, browser engines, web servers and the like, it’s as if most APM solutions try to resolve app problems by adding more hay to the haystack where they are trying to find a needle. A more pragmatic APM solution would try and reduce the haystack of data, not increase it.
Similarly, in modern engineering, you don’t have to open the hood of an engine to find faulty or corrupted parts. Today, troubleshooting complex machinery is simplified via sensors and color-coded status indicators lights.
We could simply borrow this concept for application troubleshooting, but until now, critical end-user application support has suffered from approaches that solely address the wrong audience – the application developers.
It is high-time to alleviate the complexities in APM solutions by making them work like the intuitive sensors that exactly pinpoint offenses and eliminate guesswork. This calls for high-level dashboards with status indicators that provide an end-to-end view of the app ecosystem for quick diagnosis.
Application support utopia is analytics-oriented, not metrics-driven
Paradoxically, in contrast with the design justification of most APM toolsets, almost all organizations allocate less savvy support analysts to the problem-finding tasks.
These first-line support teams mostly tend to have rudimentary understanding of application technologies and programming languages used in creating cutting-edge user experience-based apps.
This causes the support teams to seek help from experts – the developers and middleware specialists – on many issues, unintentionally wasting their precious time in app maintenance and support, rather than developing that next cool feature that will ensure end-user loyalty.
Furthermore, the rate at which new programming languages and frameworks are releasing is faster than ever, and this puts a lot of pressure on developers to update their skills, let alone the support analyst.
Demystifying application management
APM toolsets must aim to demystify application management by providing simple and intuitive screens backed by strong filtering capabilities that allow flexible pivoting and easy root-cause analysis. Fundamentally, this demands a shift in making APM toolsets less metrics-drivenand more analytics-driven.
This can be achieved by redesigning APM solutions that were never purposed for support teams and involving direct feedback from support team members. It’s critical to making APM toolsets easier to use, and also to reinforce support functions and meet the expectations of customers that invest heavily in these technologies.
APM is no longer a tool just for the experts but a technology that requires simplification to help support teams make application monitoring easy, proactive and exciting. This is the only way to empower support analysts to handle a majority of issues themselves and to minimize the unnecessary involvement of busy developers.

Blog written by Nik M Jain, principal consultant for APM Presales at CA Technologies.