Definition
Too often, performance monitoring of the enterprise infrastructure is limited to the lower tiers of the seven-layer network stack. With respect to analyzing performance, the natural inclination is to focus on the servers, routers, switches, and other equipment that comprise the physical network architecture, as well as the protocols, links, and circuits that allow these components to communicate. What’s often overlooked, or treated as a separate issue, is application monitoring.

Regardless of how well the physical network is planned and managed, overall network performance will suffer if the applications that are hosted on the network and transmit information across the network are compromised. Examples of adverse application performance include erroneous execution, which can completely disable the application or cause it to overwhelm servers or circuits, and inappropriate or unanticipated utilization, such as when personnel download large files or streaming video or when demand for access exceeds projections.

Applications
Application performance management is essential to monitoring and maintaining the health of applications throughout the enterprise. To ensure application availability and the achievement of service level agreements (SLAs) with all relevant parties, IT personnel must have their fingers on the pulse of who is using the application, when they are accessing it, where they are located, what they are doing, and much more. Application performance testing and other management capabilities become even more critical as application complexity grows in N-tier environments, as the number of users rises, as applications and their associated traffic are segmented into distinct classes of service so they are prioritized to meet SLAs and deliver a positive quality of experience for users, and as plans are formulated to incorporate new applications into the architecture.

Key Considerations
Application performance management must not be conducted in a vacuum. Passive server-side analysis and in-depth WAN- and LAN-based analysis are required to address performance problems or deviations from established baselines. When application monitoring of current and historical conditions is integrated with similar capabilities for the physical infrastructure, issues can be identified and quantified rapidly, and their source isolated to the network, the server, or the application itself.

Poor application performance produces negative effects that can be felt across the enterprise – from reduced employee productivity and IT inefficiency to customer dissatisfaction and loss of business. Application performance management solutions can help proactively avoid problems and mitigate their effects by facilitating fast, accurate reactions when anomalous conditions arise, but only when the solutions address the following key considerations:

• Application baselining and availability (regularly measure nominal performance horizontally across the entire distributed infrastructure and vertically across all of the seven-layer network stack as the operating environment evolves over time)
• Round-trip time measurement (determine how long it takes for application traffic to get from source to destination and back)
• Application tracking by traffic class (identify how much bandwidth applications such as voice, video, Web, FTP, and streaming are using, in real-time and in the past, for each class of service to promote maximum network efficiency)
• Recognition of incorrect class of service settings (ensure applications aren’t initiating performance issues because they have been misconfigured due to an erroneous class of service assignment)