This chapter discusses identifying and collating raw data required for network design and techniques for presenting those data for analysis. Capacity planning lays the foundations of the design phase; if the foundations are bad, then cracks will quickly begin to appear in the design and the costs of repair will start to spiral. It is surprising how many commercial network operators simply throw more technology and money at the problem in an effort to make that problem go away, rather than start with a solid capacity plan. In order to perform useful modeling and produce an effective design it is clearly important that the initial traffic assumptions and estimates you make are meaningful and have at least some resemblance to what is likely to happen on the network. Since these data may also be used to predict future expansion, it is even more important that these data are a reasonable reflection of reality.

Having said all this, nobody ever claimed that capacity planning was easy. In designing a new network, a considerable amount of theoretical and empirical test data must be gathered, sanity checked, and analyzed. More often than not there is already a network in place, and the proposed new design must migrate existing users and services. On a large existing internetwork with many sites and many services this task can be daunting. This task is time consuming, calls for good communication and coordination between interested groups, and requires skilled staff and sophisticated traffic capture tools. We should...