Abstract: Information stored at a site remote from a user (1) is accessible by the user via a communications network (4), such as the Internet or an Intranet. In order to provide a degree of fault tolerance, the site is constituted by a number of similar sub-sites (5a, 6a, 7a, 8a, 9a; 5b, 6b, 7b, 8b, 9b etc), which each have the same network address as far as the user is concerned. A user, who requests the supply of particular information as part of a URL (Uniform Resource Locator), is initially connected to one sub-site having a respective server (8a; 8b; 8c) and a login procedure executed by a daemon process (7a; 7b: 7c). In the event of certain faults in the one sub-site the user is connected to a second sub-site without any requirement to repeat the login procedure, as a result of all sub-sites being provided with all user login details, which are stored by session daemons 9a, 9b, 9c, and the original request is complied with at the second sub-site.