In decision analysis, there are several problems with the assignment of precise numbers to decision components, such as probabilities, values and weighs. These can very seldom realistically be estimated. Therefore, various alternative approaches have been suggested over the years, such as interval, capacity and ranking models. The more general of these are however problematical from several computational viewpoints and in this article, we deal with the server-side issues when converting the application from a stand-alone PC program to a server-client decision analytical software. On a server with a large number of users, space requirements become paramount as opposed to a single user on a PC. On a PC, matrices can be explicitly stored in memory, while on a server, to save space, matrices might have to be stored in an implicit (compacted) way, leading to space-time trade-offs.