Today, enterprises rely heavily on applications for nearly all business-critical processes. These applications are delivered via a combination of hardware, software and services, known as the application delivery chain.
To make this delivery chain work together effectively, IT must take a new and converged approach to network performance monitoring and application performance management. Let’s explore why network performance monitoring alone is insufficient.
Applications are delivered via a complex application delivery chain
Network and application performance issues are growing dramatically in importance for a variety of reasons (complexity of data centers, number of devices, rising end-user expectations). Consequently, IT organizations need a detailed, quantitative understanding of whether their applications effectively meet their business objectives.
Why network performance monitoring is insufficient
Network Performance Monitoring provides base-level reporting on network operations such as delay, packet loss and throughput. While monitoring network performance is important it is not enough on its own to fully address application performance challenges.
To fully understand the performance fluctuations of business applications and end-user experience, it is important to have an accurate representation of what network issues compromise application performance and create poor end-user experiences.
Many NPM solutions manually identify applications and users as discrete IP addresses on the network. Data on throughput, latency, jitter, and dropped/errored packets will then be ascribed to applications and users in an effort to infer application performance. Given the number of applications, users, transactions and their frequency of change, this manual approach comes with obvious scaling and accuracy issues.
While providing context to the application is easy to imagine, it is much more than simply pairing network addresses to application types and relying on network data as a proxy for application performance. To provide an accurate picture of the impact of the network on the application and, more importantly, on the end-user, specific network and application metrics must be correlated.
Mapping NPM data to individual applications provides basic reporting on how the network is performing for these applications but does not extend to what the end-user is experiencing. Although this approach provides a relatively accurate view in isolation, traffic inference is limited to the least common denominators of application performance. Specific application behaviors might not be detected, resulting in overlooked performance bottlenecks.
A broken view of end-user experience
In the typical network environment, the end-user is a constantly moving piece in the overall network management puzzle. NPM attempts to provide context to the end user by realizing that each component of the network affects the others.
The network and associated infrastructure is the most pervasive part of the delivery chain and only exists to serve and support applications. With the increased complexity of applications and the delivery chain, network and performance metrics need to shift from the traditional infrastructure (speeds and feeds) to the applications and context surrounding them.
The problem is that these traditional metrics are concentrated on the first tier, meaning that any back-end interaction or third-party connection is not included in the overall application definition and reporting.
This provides a broken view and missing pieces in the end-to-end view of true end-user experience, because it is purely based on network performance statistics and not the true experience delivered by the application.
While NPM metrics are increasingly geared toward applications, they are still displayed in the context of disparate conversations, and the performance data is not naturally correlated to end-user activity. As a result, NPM on its own provides something of a disaggregated picture of the overall end-to-end view.
As the performance of the network and the applications it supports deeply intertwine, the market is demanding a new and holistic approach to monitoring and managing performance.
To learn more about real-time visibility into complex network problems, visit http://ow.ly/dIZjK