Scalable Definition

Scalable refers to refers to the situation in which the throughput changes roughly in proportion to the change in the number of units of or size of the inputs. It can also be looked at as the cost per unit of output remaining relatively constant with proportional changes in the number of units of or size of the inputs. Scalability refers to the extent to which some system, component or process is scalable.

Throughput is the amount of work that can be performed or the amount of output that be produced by a system or component in a given period of time. It has a meaning similar to that of capacity, and the two are often used as synonyms.

Industrial processes are often described in terms of their scalability. For example, if blast furnaces can be produced in a wide range of sizes and if larger sized blast furnaces can produce proportionately more steel of the same quality and at the same cost (and smaller blast furnaces produce proportionately less steel of the same quality and cost), then the blast furnace technology is scalable.

Some industrial processes might not be scalable, or scalability might only exist within a certain narrow range. That is, there is a certain optimal size of the equipment or plant at which cost is minimized and/or quality is maximized. For example, a nuclear power plant is scalable only within a limited range, in part because because there are large fixed costs (i.e., costs that do not vary with the size of the facility or the level of output) for equipment which must be maintained for radiation containment facilities, cooling and security. Thus, it would be prohibitively expensive in terms of the cost of kilowatts of power to produce a small plant just to serve a single town or village.

Scalability is different from economies of scale. This is a term used by economists to describe the situation in which the average cost per unit of output decreases as all inputs are increased in equal proportions. Economies of scale usually exist for only a limited range of output and then the average cost begins to rise as output increases further1.

Scalability is very important to computers and communications systems. For example, computer networks that use the TCP/IP (transmission control protocol/Internet protocol) protocol are highly scalable in that overall throughput can increase roughly proportionately2 to an increase in the number of hosts (i.e., computers on the network). This is because TCP/IP itself is highly scalable. The advantage of scalability of networks is the ease and low cost of adjusting them to the size required. This scalability has been a major factor in the success of the Internet.

Some computer operating systems are also highly scalable. Among them are Linux, which can now run efficiently on anything from a wristwatch3 to a supercomputer. Its outstanding scalability is largely the result of the upgrading or replacement of a number of algorithms and data structures in versions 2.5 and 2.6 of the kernel (i.e., the core of the operating system) in order to accommodate the huge amounts of memory and hard disk drive (HDD) capacity that are required for the largest of modern enterprise-class systems.

Scalability is a key consideration for enterprises and other organizations when making investment decisions, including regarding computer hardware and software. It is important when selecting an operating system because it allows organizations to be able to grow without having to change the operating system, which can be a very complex and costly endeavor. It is also beneficial because it allows a wide variety of computers and other equipment to use the same operating system, thereby permitting use of the same application programs and facilitating the exchange of data among them.

1The exception is a natural monopoly, which is an industry (i.e., a type of good or service) for which economies of scale exist for the entire range of output that the market can absorb. But even for natural monopolies, it is quite likely that there would eventually be some level of output at which average cost would begin rising again if the market (i.e., demand) were sufficiently large.

2The word roughly is used here because this might result in increased congestion on some portions of the network. But it might also create additional network effects, which could offset the negative effects of the congestion. Network effects are the externalities (i.e., benefits) accruing to users of a product as the number of users increases, such as telephones becoming more useful as more people have them and thus more people can be called by any user.

3IBM developed a special wristwatch on which Linux can operate in order to demonstrate the great scalability of this operating system. For more information on this watch, see IBM's linuxwatch page.

Created March 7, 2006.
Copyright © 2006 The Linux Information Project. All Rights Reserved.