Case-based reasoning offers a novel approach to develop knowledge based systems. A case-based system (CBS) stores problem solving expertise as cases in its casebase. A case captures a problem description and the description of a solution to the problem. A CBS solves a problem by starting with an approximate solution found in a case in its casebase. When presented with a problem, a CBS analyzes it to extract salient features relevant for problem solving. It searches the casebase to identify cases with similar features. All such cases are retrieved and compared with the problem to select the best matching case. The solution in the best case is adapted to develop a solution to the problem. The proposed solution is evaluated. A new case is formed by combining the problem with the proposed solution. This case, if found suitable, is stored in the casebase. A CBS, thus, augments its casebase with new cases as it solves new problems. Case-based reasoning has been used in a wide range of application domains to develop problem solving and advisory systems. A limitation of these systems is that they lack adequate data management support for casebases. Most current CBS are small memory-resident systems. They use small casebases, which are loaded into primary memory during processing. This limits the size of the casebase and restricts the scope of the CBS. Since a CBS develops a solution by starting with an approximate solution from its casebase, its problem solving ability depends to a great extent on the variety and number of cases available in its casebase. It is more likely to find a closely matching case for a given problem in a large casebase compared to that in a smaller casebase. A CBS, therefore, needs a large casebase to operate at an acceptable level of expertise. As a CBS solves new problems, it adds new cases to its casebase. Thus the casebase keeps growing with the daily use of the system. A major research issue confronting CBS research is how to create large systems that can handle large casebases comprising hundreds and thousands of cases (Kolodner 1993). Our research addresses this important issue of providing data management support to large casebases.