Unfortunately, the methods of increasing effective high-priority bandwidth mentioned above have undesired side-effects. First, simply increasing the total available bandwidth does not necessarily increase the amount available to high-priority applications. In a sufficiently loaded network, for example, when more bandwidth is made available, all of the network applications sharing that link will seek to consume it and ``keep the pipe full,'' in accordance with the network protocol that has evolved. In other words, ``as soon as bandwidth is increased, it is consumed'' [5], and there is no guarantee that mission-critical applications have received their share of the extra bandwidth. In short, attempting to solve shortages by merely buying more bandwidth (without regulating it) is often a waste of money. Second, banning particular Internet connections, while it may free up more bandwidth for high-priority applications, defeats the intent of the Internet as a global communication network. Moreover, computer users on a network that bans access to some parts of the Internet are likely to find ways to circumvent these bans. By contrast, users on a capped network will be less inclined to ``bypass the system'' since their connections still work. To a large extent, bandwidth capping provides a cost-effective, guaranteed quality of service without shutting users off from parts of the global Internet. Implementing a bandwidth cap merely requires access to the router interface and knowledge of router configurations. Furthermore, users on a capped network are able to do everything that they can do on an unregulated network; the performance of certain connections suffers depending only on the amount of existing network traffic.
However, while bandwidth capping provides a more effective solution than the alternatives, it too has limitations. Two of these limitations are: (1) bandwidth caps are only as useful as the rules used in router configurations, and (2) network performance can be hindered by more than just the available bandwidth. First, bandwidth caps are only as effective as the priority classifications made by network administrators. Unfortunately, deciding which connections or computers are high-priority is often difficult because organizations rely on a wide variety of network applications that frequently change. Consequently, using router configurations to differentiate high-priority and low-priority network traffic in a connection-based cap (or high-priority and low-priority hosts in a host-based cap) requires continual maintenance and can be time-consuming. Second, even with bandwidth caps in place, other factors out of the network administrators' control contribute to poor network performance. Bandwidth capping can only control the flow of traffic between a network and the Internet. As a result, any other network delays between this link to the Internet and the final destination will cause the performance of even uncapped applications to suffer. Nevertheless, despite these limitations, bandwidth capping offers an adequate solution to organizations looking to minimize the adverse effects of over-using a shared network link.