One of the great challenges of network administrators is the monitoring of user experience. It’s become something of a buzzword or worse a cliché, especially in the hands of some who insists the network team improve it without any consensus of just exactly what “it” is and specifically how to implement the required enhancements. Without clear direction and ultimate objectives, it’s difficult to know where to begin or appreciate when as the network engineer they have actually arrived. Definitely a recipe for frustration for all stakeholders involved: customers, IT staff, and management, among others.
So let’s begin by providing a framework to have an intelligent conversation and subsequently provide actionable information—in the form of generally accepted metrics—that will define what user experience means, enable network engineers to begin measuring it, and most importantly offer specific actions that enable them to realize the mandate of increasing it.
First some background information. User experience (also referred to as end-user experience) is really about viewing the applications and services delivered from the perspective of the customers using them, wherever they may physically reside or regardless of the access method. Often for IT, this can be somewhat of a paradigm shift, given the perspective frequently eliminates nearly all the interesting speeds and feeds from the conversation engineers enjoy talking about but that users don’t see. An impressive implemented environment with the latest in VMware virtualized infrastructure, a state-of-the-art 10 Gb networking backbone and a beautifully architected multi-tier web based front end may now be in place. None of it makes any difference or is particular important to the user of the services if it does not enhance their experience.
Now let’s consider how to accomplish the goal of improving the user experience. Recently, IT analysts at NetForecast conducted a survey of 364 IT managers. Based on the survey results, they highlighted five important metrics that performance monitoring solutions should look to in order to produce a clear picture of the user experience.
Metrics and Monitoring
The first metric is the end user’s page response time. This is a measure of the time required for the original request to be processed. It could be measured by placing a probe near the client to measure the turn time and validate the processing of the request. An example would be the duration from a request for a website and when the content is displayed on the user’s client.
Next is the total number of transactions processed. During a specified time period, measuring the volume of transactions is useful to see if they are too high, which results in transactions being caught in the queue. This in turn leads to errors and causes issues such as the client processing the request again. Excessive transactions can materially impact individual user experience.
Server query response time is also an important metric. This is the application-server side counterpart to page response time. Application-specific visibility is necessary in order to assess detailed response times and view transactions. Continuing with the website example, this might be the time the web server waits until it is able to construct the entire URL contents.
Another advantage of having application-specific visibility is being able to measure traffic flow data. Data can be collected about each conversation showing the flow by application, including such details as packets, bytes, connections and request details. Understanding the traffic flow mix and amount on a network can provide invaluable indirect information into how distinct users may currently perceive usability. Perhaps more important, it can provide insight into trends that may eventually impact users if remedial action is not completed.
Finally, server errors themselves provide useful information. Details of the history of packet captures and application transaction details show server conditions, and this visibility shows when errors result from a higher number of requests than the server can handle. Too many server errors can ultimately impact users’ overall service delivery experience.
Network latency (RTT), server utilization, network availability and bandwidth utilization are additional metrics that provide insight into the user experience. Leveraging to many of any of these or in combination with the above discussed parameters can impact the delivery of services to users.
Employing Best Practices in Monitoring
A large part of the picture, in addition to monitoring the application data and general network conditions, is the infrastructure supporting the system. Administrators can enhance their visibility and awareness of underlying resources by seeking a solution that will provide several key features in monitoring the network and proactively managing traffic.
- The underlying components that support user applications should be monitored to ensure proper application delivery. These devices serve as the foundation of the entire delivery process and must be running at optimal efficiency to maximize overall service delivery.
- Baselining gives IT a long-term view and a starting point to assess when performance is higher or lower than expected. This is an often overlooked capability that can provide valuable insight into behavior over extended periods of time. In particular, observing activity over time windows greater than six months can yield slowly degrading performance that might otherwise be missed.
- The user experience isn’t for end users alone. IT should have a single dashboard through which network activities can be monitored and managed, providing an at-a-glance problem identification. This perspective can be the starting point of any potential troubleshooting that may be required.
- Video performance is increasingly important as telepresence calls replace face-to-face meetings. Network administrators should validate unified communications performance, including VoIP analysis, to ensure satisfied users. Of particular importance is to understand how the significantly higher utilization in conjunction with high levels of network priority may impact all applications and services operating on the infrastructure.
- Resources can be used more efficiently if monitoring activities are conducted entirely by the network probe, which reduces overhead. Therefore, seek those solutions which perform the majority of their processing locally at the probe.
As the above conversation has shown, quantifying users’ perception of the services which they are utilizing is vital for IT to fulfill its mission. This observation does not so much change what a network engineer has been doing, but instead measures these activities from the perspective of the customer’s point of view. By constantly asking if the decisions and actions being undertaken will improve the user experience and validating these assumptions by the metrics described above, a network engineer can be assured of providing the optimal services for their IT investments. The right application, network, and application monitoring solution can help achieve this objective, making the most out of bandwidth and computing resources, reducing expenses and more easily resolving issues that arise to keep the business running smoothly.