Cost in computing can generally be defined in either financial or technical terms. With respect to financial cost, some typical factors for budgeting include:
The combination of these and other factors is often called the Total Cost of Ownership (TCO).
Although in accounting theory, any operational factor can be weighed with respect to its estimated dollar cost, in many cases other more technical metrics of computing performance are more direct and useful. Technical metrics such as these include:
In algorithmic theory, the essential cost dimensions of a process are the number of operations required (time) and amount of data storage needed (size). The theory of algorithmic complexity provides order-of-magnitude estimates of the comparative time and space requirements of various alternative solutions to problems such as searching, sorting, and routing. Dijkstra's algorithm for the computation of a least cost path spanning tree from a weighted graph is an example of an abstract cost-minimizing algorithm that is applied to practical problems in routing.
In network routing, different routing protocols use different cost metrics. In RIP the cost of routes is measured by hop count. Cisco's implementation of OSPF measures route cost in aggregate bandwidth, with higher bandwidth routes being cheaper. Cisco's EIGRP includes a metric calculated through a combination of bandwidth, delay, load, and reliability. In general, routing metrics are non-financial, although routing tables can typically be configured to minimize traffic on fiscally expensive links.