Will IT Operations Analytics Platforms Replace APM Suites?

By Graham Gillen	Article Rating:
January 12, 2013 10:00 AM EST	Reads:	1,363

Gartner recently published an important report titled "Will IT Operations Analytics Platforms Replace APM Suites?" The report is based on Gartner client Inquiries about enterprise performance and availability strategies and these enterprises report a near-universal commitment to the idea that monitoring, performance and availability should be application-centric. At the same time the report suggests the rate at which investments in new on-premises Application Performance Monitoring (APM) technology appears to be slowing down. It attributes this to observations that enterprises have perhaps prematurely come to believe that IT operations analytics (ITOA) platforms can be deployed as a substitute for traditional APM suites. It concludes that IT operations analytics platforms are not a substitute for traditional five-dimensional APM portfolios, but the importance of analytics to APM will continue to grow over the next five years. It features a strategic planning assumption forecasting that by 2015, 60% of Global 2000 enterprises will consider ITOA a higher priority than APM, up from 20% in 2012.

This report hit about the same time as a discussion on the APM Strategies LinkedIn group about the role of advanced analytics and behavior learning technologies as a priority in the Gartner APM model. So much of my thoughts on the emergence of analytics apply to both topics, and they come from the perspective of Netuitive's customers - which include enterprises in the Telco, financial services, insurance, and e-commerce industries.

The one thing all of them have in common is that they view customized applications and IT services as a competitive differentiator - not necessarily IT infrastructure, which was where monitoring started. And these companies view certain classes of applications as being critical to the business. Often they are customer-facing, revenue impacting services like Point-of-Sales systems in retail and e-commerce, or Broker Portals and Payments Systems in the financial services industries.

Each of our customers also has found that while many APM tools have some sort of analytics, it is often really just a form or reporting for use in non-real time (like capacity planning) or for post-mortem forensics. The largest and most forward thinking organizations have sought Behavior Learning technology as the unifying IT analytics across their entire APM environment enabling them to understand the business impact of IT.

What Behavior Learning technology offers is a way to learn in real time which components of your application infrastructure can be correlated to the measurable behaviors that you care about - end user experience and business KPIs. For example, can a slow-down in web page response be correlated to lengthening SQL call queues in a database intensive application? And if so, which DB cluster or DB server in particular? Or maybe it has nothing to do with the DB cluster and it can be correlated to some "abnormal" JVM metric behavior - so the app cluster is the problem. And if there is a problem - can you correlate this to business impact - like the sales per hour metric for your Point of Sale system?

As exhibited in this video - http://resources.netuitive.com/case-study-mobile-application-provisioning-with-predictive-analytics - Behavior Learning technology as a new form of analytics typically provides application performance insight useful to really deliver proactive management - and I mean REALLY proactive. The challenge of course, is that this is all foreign technology to your DevOps teams so there has to be a buy in period - especially from Sr. Management to make this work.

So from my view of the market - which is our customers - I do see very strong indication that for "APM 2.0" or "Next Generation APM", architects and APM strategists are realizing that it's not enough to collect the data and run-time architectures in the other four dimensions of APM, and that new ways of analyzing all this data - including using Behavior Learning technology - is what is needed to get the most value out of APM for your most critical applications.

The market will either have this need for analytics served by an APM vendor who greatly (and I mean GREATLY) improves its analytics capabilities, or by an analytics specialist like Splunk and Netuitive, that complements traditional APM tools.

6.5 million software developers and still going strong

Java Developer Magazine