A TCO Analysis of Cluster Servers Based on TPC-H Benchmarks

Abstract

Summary form only given. In this article the author presents a TPC-H benchmark-based cost analysis of cluster servers as computing configuration alternative to consolidated servers available in the market. The analytical model is expressed with mathematical equations to elicitat the issue and demonstrate the results, starting from Gartner's total cost of ownership (TCO) concept and factoring its costs items in terms of hardware and software costs in a server computing approach. The TPC-H benchmark data employed to test the model show that a cluster server configuration may be not a good idea in some computing scenarios, mainly in medium to large database systems, but could be best fit, economically speaking, in scenarios where are employed commodity hardware and license-free software to implement the cluster. And, reinforcing former research of other authors, the model shows that software, much more than hardware, is the key set of components in an economic analysis of a server cluster