A baseline is a value or profile of a performance metric against which changes in the performance metric can be usefully compared. For example, if you know the average utilization of a particular link over the last week, this average could be used as a baseline against which to compare future changes in the utilization of that link. A more complex baseline might be last week's actual pattern of utilization on a particular link (often called a utilization profile).

What is Baselining?

Baselining is broad term for any analysis method that compares changes in actual data against a baseline. The most common use of baselining is as a tool in performance management for trending analysis--comparing a performance metric to a historical value to find a trend that can be used to estimate future performance or needs. Another common use of baselining is in managing SLAs or service-provider agreements . In these cases actual performance is measured against the baseline of an agreed minimum service level. A third use of baselining is for monitoring network health (watching for changes in problem indicators), which is a proactive form of fault management. Although these three uses of baselining have quite different objectives, the principles (and tools used) are very similar.

Performance Management Is Not Just Baselining

The terms "performance management" and "baselining" often are thought to mean the same thing. But they are different. Performance management utilizes baselining techniques for trend-analysis purposes--to develop a picture of future requirements from past trends. But performance management is not just baselining. To effectively plan network-capacity requirements, it is necessary to take into consideration future changes in the application environment that may cause changes in the network load. For example, if a new client/server or intranet application is replacing a core-business legacy system, historical performance data is largely irrelevant because the new application will work in a different way and have a different impact on the network. In such situations it is necessary to use application-modeling techniques to determine the likely impact of the new applications.

Since baselining techniques are similar regardless of their purpose (Performance Management, Service Level Management or Fault Management), I first discuss baselining as a whole. Then I will discuss application modeling separately since this aspect of Performance Management is quite different.

Baselining

Regardless of the purpose(s) for which you are implementing baselining, there are seven steps that you should work through in developing the strategy:

1. Key service(s). What are the key services being provide to customers (internal or external) by your part of the network organization.

2. Service elements. What are the discrete elements which are chained together to provide the key service(s).

3. Metrics. What key services or service elements actually need to be measured and what measurements are required?

4. Baselines. What baseline is associated with each metric. Will single values or profiles be used as the baseline?

5. Data needs analysis and instrumentation. What data needs to be collected to measure actual performance and what instrumentation needs to be added to the network to provide that data?

6. Reporting and Alarms. What analysis tools are required to collect, summarize and interpret the data so that actual performance can be compared to the baseline metrics.

7. Control. Responsibilities and procedures. E.g. Who is responsible for identifying and provisioning required bandwidth changes.

Note that this methodology is not strictly chronological -- Steps 3 through 6 are typically iterative and though Step 7 often will be completed last, it may not always.