Pricing and capacity decisions of clustered twin-computer systems subject to breakdowns

Abstract

More and more organizations are running a clustered twin-computer system to tackle the rapidly growing demand of computer capacity. For example, the German airline Deutsche Lufthansa uses two Unisys 2200/644 mainframes to handle the torrent of data from passenger reservations and check-ins, baggage handling, and flight scheduling. Basic queuing theory suggests that consolidating twin-computer systems to a bigger system with the same overall capacity will result in a lower total time in the system. The prevailing rationale for a clustered twin-computer system is that it is an effective way of coping with not only the capacity growth challenge but also the computer downtime problem. This paper examines the impact of breakdowns on the decision to consolidate or cluster computer systems. This research finds that both the consolidated single computer system and the clustered twin-computer system, under the same breakdown parameters, have the same effective computer capacity. However, the clustered twin-computer system has a shorter expected time in the system for most cases. The clustered twin-computer system also performs better when there is a heavy traffic intensity. For firms with a consolidated single computer system, this research finds that it pays off to reduce the mean time to repair than to increase the mean time between failures. Finally, this paper derives the optimal pricing and capacity of a firm's clustered twin-computer system that will maximize the net value of computing by incorporating the analytical model in a microeconomic framework to consider the effect of delay cost of computer processing.