Baselining is a technique used by computer network administrators for the analysis of data that has been accumulated over some period of time. Baseline data is collected periodically so that the performance of the network can be properly evaluated. The collections of data are then used as a reference during real-time monitoring and evaluation. While the concept may seem simple, it is more akin to an iceberg, where beneath the surface, a greater complexity exists.
It can be easy to confuse baselining with other network management and evaluation techniques. Often times, it can be confused with performance management, which is used more for ongoing planning. Since performance management is a continuous process, it uses the collection of data in a baseline as reference. It’s the trends that appear over time, by analyzing baselines, that are used in performance management for proper decision making when considering possible changes to the network. The truth is, the data in baselines is also used in several other aspects of network management besides performance management, including service-level analysis and fault tolerance evaluations.
A common mistake can be made in baselining by network administrators who believe the technique to be too time-consuming. In a poorly constructed baseline, the collection of data can be far too generalized, which cannot provide an accurate resource from which to gauge measurement. Without a sufficiently detailed focus, it becomes difficult to narrow down what problem areas need improvement. Simply tallying up a set of averages by asking how quick the network was each week doesn’t even begin to scratch the surface.
Once a bit of planning has taken place, the actual metrics need to be collected. Depending on the extent of the baselining or its overall goal, different data can be collected and can quite possibly consume a large amount of storage in a short time. In the case of a performance management evaluation, for example, the data may include the latency and bandwidth being used on a particular backbone router or set of branch routers. The measurements then provide the baselines that can be used for planning the growth of the network. The size of of the collected data on disk can become very large, too, when considering the accumulation of a few kilobits of information every few minutes over the course of weeks across multiple devices and interfaces.
After all of the data has been collected and organized, it needs to then be monitored and evaluated. In some situations where baselining is being used for things like capacity planning, reports are periodically generated from the comparison of the historical data with some period of measurement. Such reports can give a clear view as to when and where upgrades to the network should occur. It is also possible to have baselines continuously evaluated in real-time and alerts given out when a predefined threshold has been exceeded. In these cases, a network administrator can be given an immediate notification of a problem occurring at any particular location on the network.