The Problem of Network Design
The manner in which the network services and functions are provided is dictated by the associated design. The design, in turn, is dictated by the specifics of the service provider or enterprise network. Suppose the network core has oodles of bandwidth, such as in a fiber-based network. In such a case, the network can be over-provisioned; for example, by calculating the maximum possible required bandwidth and doubling it. On the other hand, if the provider has purchased a fixed amount of bandwidth from another provider, that bandwidth must be carefully allocated to the end users.
Naturally, these are the extremes—many networks will feature segments with plentiful bandwidth as well as segments where resources are strictly limited. All such cases require specific designs.
It’s also possible for customers to engineer the CE-PE link by marking IP packets with application-specific DiffServ codepoints. Each application or service (such as VoIP) can use a different codepoint, with the intention that the associated traffic will receive different processing as it traverses the network core.