The complexity of application delivery environments is outstripping the ability of APM products to monitor and manage performance. As a result, application managers are less interested in APM products than they were 18 months ago. That's the top finding of our Application Performance Management Survey of 300 IT business leaders about their use of the technology.
Over the past few years, the big management software vendors have been on a buying spree that's resulted in overlapping tools and poor integration, which at least in the short run has only exacerbated the problem. The level of customer frustration with APM tools is at an all-time high. Our survey found that for those companies opting out of APM tools, 51% say the products require too much staff time to implement correctly -- a jump of nearly 20 percentage points from our survey finding two years ago. What happened?
Application performance management is focused on ensuring the overall quality of application services users receive. Who wouldn't want that? In our survey, 86% said application services are very or critically important in their companies, about the same percentage as two years ago. The tolerance for application service problems has been consistently low over the past several years, though there seems to be an increase in tolerance compared with 2010. There's no question that applications are the primary interface that users have with the IT department. Our laptops, mobile devices and desktops are all provided for the goal of accessing applications. To a great degree, therein lies the problem: Delivering everything to everyone, all from a dynamic, virtualized data center, makes performance assurance much harder than it used to be.
The Complexity Of APM
In the early days of APM, software vendors provided application-specific agents that monitored not only operating systems but also critical elements of the applications themselves. Usually, some sort of network monitoring tool checked on the transport for the underlying applications. The agents focused on monitoring database applications, Web servers, email and other standalone business applications. Agents for custom applications were and are a lot harder to create, so for those apps incidents are usually tracked from calls to the help desk and then tossed over the fence for the application developers to take a look.
The underlying problem then and still today is that when users experience problems, these monitoring tools can't identify the cause. System, application and network groups often point fingers at one another, as they can't identify which component of the infrastructure is causing the problem. Countless hours spent hunting down problems are the stuff of Sherlock Holmes novels. This frustration has caused many IT organizations to seek tools that can provide end-to-end monitoring and identify root cause problems, or to give up on APM all together.
Our recent data shows that the fraction of applications monitored by an APM system is down considerably since our last survey in 2010. Despite the low tolerance for application problems and the importance of applications themselves, fewer than 40% of the companies in our survey use APM tools, consistent with the findings of our 2010 survey.
The vendor community's approach to APM is quite fractured. Respondents to our latest survey cited more than 40 vendor tools of the more than 100 on the market. The most widely used APM vendors in our survey, all cited by 20% or more of respondents, are CA, Citrix, EMC, Hewlett-Packard, IBM, Microsoft and Oracle.
download the InformationWeek February special issue on application performance management