Baselining Your Infrastructure

Networks are built by interconnecting various devices (such as routers and switches) running a variety of protocols. Protocols are configured with a number of parameters that enable them to behave in a certain way. These protocol configurations influence the traffic patterns in a network. Understanding these patterns from a unified baseline view is the first step in a successful application deployment.

The collection of baseline data is very important to the overall application deployment. The accuracy and pertinence of your baseline will dramatically affect the deployment project. Baselining is a technique used to characterize the existing IT infrastructure prior to application rollout and, subsequently, to track the application's performance.

Prior to application deployment, you must characterize existing network paths, servers, and databases that new applications will share. When these applications have their own dedicated servers, baselining of the servers' performance is the primary requirement. When the new application will reside on existing servers, however, the shared servers should also be baselined relative to their existing applications.

For the baselining effort to yield meaningful results, you must do the following:

• Collect just the data that is pertinent to the application deployment For example, ignore nonessential or nondependent functions such as name resolution.

• Collect the data at an acceptable frequency Network and application usage tends to be bursty. If you collect data points too infrequently, the bursts of activity are averaged out.

• Collect the data for an acceptably long period of time An acceptable period of time takes into account items such as month end. A typical period to cover normal aspects of an enterprise activity is 1 month.

• Determine the tolerances to design for Do you want to guarantee that a user never experiences degradation? Or do you allow for some degradation in the interest of cost?

Baselining is key to a successful strategy and forms the basis for pilot and contingency planning. It provides a constant reference point in the verification process and is used in an ongoing manner to assess the success of deployment against desired results.