App performance '1.0' bad for cloud, SOA: exec
The founder and CEO of an application performance management (APM) software vendor said that today's IT development and production environments are not adequately served by what he terms "APM 1.0."
Jyoti Bansal of San Francisco, Calif.-based AppDynamics Inc. said applications are increasingly distributed, given the movement to cloud, virtual and service-oriented architecture (SOA) landscapes. "The whole complexity and the moving parts and dynamics of applications, both in development and in production, are changing very rapidly," said Bansal.
APM tools built eight to 10 years ago -- what he terms "APM 1.0," -- "don't work at all" for modern IT landscapes, said Bansal, because they don't allow the IT admin any insight into how end users are being impacted.
AppDynamics is launching the newest version of its APM tool, with several new functions compared to predecessor, v. 1.0. There is cloud orchestration for dynamic capacity provisioning and complex policy creation in internal and external clouds. A self-learning policy engine adapts to the load and performance demands on business applications.
According to Russ Conwath, senior research analyst, with London, Ont.-based Info-Tech Research Group Ltd. there probably was good reason several years ago to term some vendor offerings as "APM 1.0," but that's changed in recent years with the availability of tools that do work for distributed applications.
Conwath said AppDynamic's offering isn't alone in working across cloud, virtual and SOA environments, although other vendors like Hyperic and Optier may not outright market themselves in that manner.
"That's the nature of APMs, in that they are intended to function in a non-homogeneous, heterogeneous environment," said Conwath. "The technology should be somewhat agnostic."
What might distinguish AppDynamic is that the software works for Java, .Net and SOA environments, said Conwath.
One of the biggest challenges that Bansal has heard from customers is that APMs are typically difficult to implement, often requiring external help be brought in.
The complexity of implementation depends on the robustness of the APM, Conwath pointed out. Having evaluated Optier's software, a higher-end product, Conwath said it might take some consulting time to set up, but that of Hyperic is more capable of scaling.
That said, Conwath has observed vendors in the last year increasingly building tools that are easier to deploy and use, even building in learning engines.
Follow Kathleen Lau on Twitter: @KathleenLau