Increased network densities found even in smaller, less complex Wide Area Networks (WANs) means that performance monitoring isn’t reserved for the enterprise administrator alone. Networks of all sizes have considerably more traffic flowing through them nowadays, potentially leaving IT managers scrambling when performance anomalies grind services to a halt.

The variety of monitoring systems and tools available make it hard to define a standard methodology for network performance monitoring. The variations of network configurations and their requirements also dictate the kind of metrics IT needs to capture.  There are however, certain considerations that IT managers and engineers should keep in mind when implementing a performance monitoring strategy.

1.    Establish your performance baseline

Network performance testing is a means to establish a requirement versus capacity ratio that determines how capable the network is to provision business tools to the organisation optimally. Understanding the overall capacity of your network requires establishing a performance baseline that measures the current, standard operating conditions of your network. Variations on either side of your performance baseline indicates changes as they occur within the environment. This aids engineers in understanding the ebbs and flows of the network and detect performance problems as they develop.

2.    Capture network traffic with Netflow

Netflow’s (and some variations thereof) popularity makes it a good monitoring protocol for capturing flow data from routers of various brand names. Flow data is sent to a Netflow Collector that stores and manages data for analysis. Adding to Netflow’s popularity is the non-intrusive way it collects packets from network devices and keeping processing demands on routers to a minimum. Netflow data can give rich insight to application data usage, bandwidth bottlenecks, router performance and health statistics. It also has flow aggregation features that allows it to combine multiple data streams to represent holistic flow information for accurate network reporting.

3.    Don’t confuse productive utilisation for bottlenecks

As you proceed to sample performance metrics, it is important to view your findings within the context of the time-frames they have occurred. All businesses have peak productivity phases that may include accessing a certain application or other resource at a certain time of the week, day or month. Finance departments will typically grind the payroll system in the last week of each month. The payroll system in turn, might query HR application server to correlate employee information, causing a spike in network traffic. It is crucial for IT managers to have an awareness of productivity cycles such as these to understand the true nature of its performance fluctuations.

4.    Deconstructing application traffic

Network reporting and monitoring tools are capable of deciphering network traffic down to granular detail, allowing engineers to distinguish between, for example, web browsing and cloud-based application traffic. This gives you the ability to determine usage volumes and patterns between applications in a very detailed manner, and also draw on the metrics to find ways of addressing specific problem-applications and eliminating guesswork in the process. With so many application traffic traversing the environment, your network performance testing tools should give you granular views that elucidate performance concerns, not make them more complex.

The importance of a network performance testing regiment

Network performance testing isn’t one of those “set and forget” tasks. To leverage the true benefits of network performance tests, you must do it on a continually evolving basis. If changes occur within the environment, your initial baseline no longer serves its purpose. With change, comes varying degrees of demands for network resources. IT staff should continually evaluate their capacity in realistic terms by constant evaluation of their networks.

Image Credit: Mrkchas

.