Network Performance

Put monitoring strategy on offensive

Understanding bandwidth and resource consumption is the key to better network performance management.

by Dirk Paessler

It the most basic level, accessing network data and putting that data to work through a variety of applications is what businesses do all day, every day. Basic communication and mission-critical operations—including everything from order entry and customer support to supply chain management and employee benefit administration—all rely on network data and moving that data between sources.

A network comes with a fair share of management headaches. Usage spikes can reduce application performance to a crawl; too little bandwidth can cause Web site visitors to become impatient and leave before completing purchases or finding data they need. Internally, frustrated workers can create additional load on IT departments and detract from strategic initiatives. Monitoring critical network resources is, therefore, an essential task for every organization.

Fast servers and more bandwidth would appear to be the solution to maximizing business processes. Achieving peak network performance, however, is not that easy.

In most cases, once issues are detected, network performance has already been compromised. Instead, network administrators and IT managers need to be on the offensive with their network-monitoring strategies, employ continuous monitoring to diagnose problems and resolve them in real time, before they have a serious impact on business.

The source of network slowdowns can be the network itself, limited bandwidth or a bottleneck at the application level. Analyzing various aspects of the network, including packets and flow-monitoring data, can make identifying the sources of the problem easier.

The first decision in creating an effective monitoring strategy is to determine exactly what aspects of the network to monitor. This will impact the type of monitoring software and what methods to use. Typically, administrators monitor hardware or application failures on servers, resource usage by servers, disk space, CPU and memory usage on servers and workstations, and network bandwidth usage.

Implementing a monitoring solution alone is not enough to effectively monitor networks. Focusing on the network's vital signs also is important, including availability, speed and usage. Most aspects of network performance can and should be evaluated in terms of these three criteria–turning data into performance metrics and answering the question of what change, if any, can be implemented to improve performance.

Monitoring software should be easily configured and integrate with other network-management applications, if needed, and be easy to use. Monitoring software should also offer good application discovery and identification, and in-depth views of user experience. Presenting real-time and historical information provides greater insight into what is happening on the network.

There are a number of steps than can be taken to diagnose problems easier, faster and more accurately.

Track key performance indicators and trends. Look for variations in traffic load and traffic types, as well as variations in application performance. Determine what is causing spikes, dips and changes. Are these variations planned and necessary? More importantly, is the network able to scale to handle the variations?

Some bandwidth strain is planned and necessary, and can be managed to soften its impact on performance. Scheduling and performing data backups during off hours is one example. If the backup is taking longer than expected and reduced network speed is affecting employee productivity, consider adding faster servers, tape drives and network switches to enhance performance and balance the added resources. The key is to anticipate resources before they negatively impact performance.

Identify security risks. Traffic and load variations can indicate a security breach or increase vulnerability. Traffic spikes in a certain part of the network, for example, can be a sign of a malicious worm that spreads from one rogue PC. Spikes in traffic accessing a Web site can also be a denial-of-service attack.

Keep a close watch on high-risk applications that hog bandwidth. Certain applications and configuration issues consume excessive bandwidth and cause bottlenecks. Peer-to-peer applications, VoIP applications, instant messaging software, continuous streaming video and e-mail with large attachments are common culprits. The use of unwanted protocols and incorrectly configured connections between servers and clients, as well as between infrastructure devices, can also cause bottlenecks, performance delays and denial of service.

Reduce and prevent unwanted traffic. Create and enforce acceptable use/usage policies that define access parameters and discourage high-risk activities or the use of applications, such as streaming video, that consume excess bandwidth. Communicate the policies broadly. Employees may not even know how their activities are impacting others on the network. Building awareness of common issues and guidelines is imperative to optimizing overall network performance.

Understanding bandwidth and resource consumption is the key to better network performance management. Network administrators can avoid bandwidth and server performance bottlenecks, plan upgrades of infrastructure strategically and deliver improved quality of service to users. Network traffic analysis can be used to support the business case for implementing IT change—adding bandwidth or hardware, deploying server load-balancing solutions and implementing best practices.

Continuous network and server monitoring enables staff to find problems and resolve them before they become a serious threat to the business. By tracking and documenting short-term changes in network performance, as well as long-term usage trends, network administrators can implement strategies and best practices that not only diagnose network problems, but help mitigate risks before they occur and optimize overall network performance.

Dirk Paessler is CEO of Paessler AG, Germany.

For more information (click here)