Use of virtual machines in information systems

Abstract

This paper presents a scheme using the virtual machine concept for creating: 1) An environment for increasing the effectiveness of researchers who must use analytical, modeling systems and have complex data management needs; 2) A mechanism for multi-user coordination of access and update to a central data base; 3) A mechanism for creating an environment where several different modeling facilities can access the same data base; 4) A mechanism for creating an environment where several different and potentially incompatible data management systems can all be accessed by the same user models or facilities.

The paper investigates and formalizes the performance implications of this scheme specifically directed at the question of response time degradation as a function of number of virtual machines, of locked time of the data base machine, and of query rate of the modeling machine.