Right-sizing computer resources for re-engineered processes

Business Process Re-engineering (BPR), with its emphasis on empowerment, productivity enhancement, and supplying decision-makers with timely information, is strongly correlated with the migration from centralized, mainframe to distributed, client/server (C/S) computing environments. The concomitant reduction in the MIPS cost associated with the smaller host-servers in C/S environments has made it possible to provide multiple hosts in the C/S environment to address issues of security, data replication, and the host-server as a single point of failure. The provision of multiple host-servers in the C/S environment also permits service replication to provide optimal response times and handle transaction loads during peak On-line Transaction Processing (OLTP) periods. While multiple host-servers successfully address the problems mentioned above, their presence impacts the assumed cost savings of C/S computing environments. For this reason, it is increasingly important to select host-processor sizes that are appropriate to the services provided and, in the case of replicated services, to balance the replicated services over host-servers to achieve the optimum use of computer resources. This paper addresses these latter two issues and specifies computations and load-balancing mechanisms, including automated ones, that will assist managers in determining the “right” size and number of host-servers needed for a particular re-engineering effort