The duplication of data on a number of computers within a distributed system. There are a number of reasons for duplicating data. The first is that computers malfunction and go out of service; if only one computer held a single copy of an important set of data, then this would become immediately unavailable. Data replicated across a number of computers eliminates this problem. A second reason is to increase access time. Data can take a long time to travel over large computer networks, particularly wide area networks. In order to achieve fast response times such data is often replicated in computers that are close to those which access it: for example, the data might be contained on a computer which is connected via a local area network. Data replication is not free. One of the disadvantages of replicating data is that each of the files or databases containing the data must be kept up to date. This means that a network often has a large number of update messages flowing through it as users interact with the data, modifying and deleting records. Designing a distributed system containing replicated data so that it is reliable, achieves a good response time, and is not swamped by update traffic is still an art. Data replication has achieved a major importance with the increasing use of network technology. There are now a large number of replication utilities which are either built into conventional database management systems or can be purchased as stand-alone programs. See also mirror site and mirroring tools.