Building an app-centric Infrastructure Performance Monitoring system

How to continuously capture, correlate and analyze system-wide heterogeneous infrastructure data so you can compare it to established response time, utilization and other metrics

code script
Thinkstock

This vendor-written tech primer has been edited by Network World to eliminate product promotion, but readers should note it will likely favor the submitter’s approach.

Applications are the lifeblood of any enterprise, but application performance can suffer without optimal support from the infrastructure teams. At the heart of this problem is the fact that application- and infrastructure-layer management stacks remain distinct entities across the IT environment.

Application performance management (APM) drills down into the behavioral characteristics of the app and its supporting runtime environments, and perhaps select pieces of the virtual servers, but it is blind below the virtual machine. The underlying physical and virtual resources are managed and controlled by an infrastructure performance monitoring (IPM) system, which can effectively maintain the health of physical and virtual resources but doesn’t know whether this is producing an effective application environment in terms of either cost or performance.

Neither toolset can provide a fully integrated performance optimization solution, which is forcing the two areas to converge.  Here’s what IT teams should keep in mind as they bring together their application, web, database and infrastructure teams.

Achieving synergy between applications and infrastructure is more than just blending disparate management regimes. A functioning application-centric environment requires enterprise executives to make changes to their current ecosystem on both a systems and an operational level. This can be difficult for organizations that maintain substantial legacy infrastructure geared toward conventional data workloads.

One of the first things to do is to stop depending on silo-specific tools. When application requirements were fairly predictable, it was common for organizations to provision infrastructure to support the most demanding circumstances, even if that resulted in over-provisioned resources that would sit idle for long periods. This also often led to isolated application and infrastructure environments within the datacenter ecosystem as solutions were crafted to solve unique challenges at particular times.

But this is simply not tenable in a modern application environment where traffic is unpredictable, resources are often beyond the direct control of the enterprise, and competing products can come out of nowhere to upend established business models. Modern infrastructure needs to be flexible, rapidly scalable and provide a high degree of federation across resources and applications.

Most enterprises have made great strides in de-emphasizing silos through virtualization and the transition to software-defined architectures, but there is still a substantial portion of infrastructure sitting on traditional hardware. Sometimes bare-metal support is crucial to application needs, but more often than not it merely creates a barrier to agility and flexibility. It is also important to note that silos are not just a technology problem, but a cultural one driven by the rise of digital fiefdoms within large organizations that usually prefer to hoard data rather than share it.

Silos are particularly detrimental to efforts aimed at fostering Big Data and the Internet of Things. Without the smooth flow of information across the entire enterprise footprint, analytics performance will suffer and the results drawn from incomplete data sets will, by nature, be inaccurate.

Organizations should also strive to achieve real-time performance across their entire data center infrastructure because the pace of business in the digital era is nearly instantaneous. If the infrastructure performance management stack cannot function in real-time, neither can the applications – at least, not in ways that are beneficial to broader strategic goals.

By building real-time performance visibility as a core asset, the enterprise can address performance issues before they inhibit the user experience and support continuous development and optimization strategies to strike the right balance between resource consumption, performance and a host of other factors. 

An effective application-centric IPM approach will help teams continuously capture, correlate and analyze system-wide heterogeneous infrastructure data and compare it to established response time, utilization and other metrics. To do this, it must incorporate an advanced analytics framework capable of providing a contextual understanding of the application environment and enable a robust set of correlation, discovery and predictive data analysis capabilities on an intuitive dashboard. It should also provide vendor-agnostic insights that are accurate and actionable by engineering and operational teams, and it should scale to appropriate levels without sacrificing speed or functionality.

A third element in the app-centric IPM environment is the balance between individual user requirements and those of the broader community. Users are increasingly demanding customized, even personalized, services, so it is important that the enterprise can fine-tune the performance of individual apps and workloads – but this must be coordinated with a broad view of conditions across the infrastructure. Without this multi-layered approach, issues like improper load balancing and resource contention can flare up at a moment’s notice, even on abstract, virtualized infrastructure.

With all the changes taking place in the business world today, one principle remains constant: a company is only as good as its reputation. In a service-driven economy, trust can be built very quickly, and then just as quickly destroyed.

Organizations that hope to compete at this level need to ensure that the infrastructure that supports their application environment delivers rock solid performance and availability at all times, even as it continually evolves in order to tap into emerging markets.

With an application-centric IPM approach, the enterprise can leverage technology to focus on what really matters: enhancing the user experience and driving revenues.

To learn more visit Virtual Instruments

Join the Network World communities on Facebook and LinkedIn to comment on topics that are top of mind.
Must read: 10 new UI features coming to Windows 10