The relationship we enter into with our network equipment suppliers can, and sometimes does, resemble a marriage. Indeed, some of us might be guilty of keeping our eyes open before the marriage, and half shut afterward, as Ben Franklin used to say.
Our vendor “partners” got busy churning out products that we burned money on in order to plug into our enterprise networks because we bought into their plans. Left out of the equation was any internal analysis of what problem the product solved or what specific network conditions we sought to improve based on performance measurements. Network infrastructure became a moving target because of these rapid deployments, and we amassed little real in-depth network knowledge.
Given the current economic climate, vendors no longer can push product on companies. Enterprise network executives must assess their immediate and long-term needs, and vendors must earn their way into the network by providing the “right” solution.
That’s no small feat for network executives. It means rediscovering your network, knowing what you are looking for in products and understanding how to determine if it exists.
Rediscovering your network means you’ve got to invest some time understanding its operational characteristics. How many errored frames are common in your transmissions? What are the network choke points? Where and when do you experience spikes in usage, and how does the network respond? What are your application characteristics, and can the network support them adequately?
In this new discovery mode, structured benchmarking is essential. The structured approach is applicable to any technology, and the aim to isolate specific elements of the network and pinpoint performance. But to do that with success, you’ll need to follow a handful of basic principles:
• Establish testing goals. It is important to identify the devices to be tested, but more so, identify the trade-offs between what data is “doable” to collect and what is desirable.
• Implement a test design methodology. Here you need to profile the major elements to be tested and develop/determine the appropriate criteria for evaluating them.
• You’ll need to conduct prototype testing and validate your results. This means building a viable test bed and creating prototypical tests to ensure that results are meaningful and insightful.
• Production testing comes next. Here you capture, compile and record raw data.
• Finally, you document your data and analyze it.
Tolly is president and CEO ofThe Tolly Group.Reach him via e-mail at ktolly@tolly.com.
