Submitted to the System Design & Management Program In Partial Fulfillment of the Requirements for the Degree of
Master of Science in Engineering and Management
MASSACHUSETTS INSTITUTE OF TECHNOLOGY February 2007
© 2007 Ken Huang All Rights Reserved
The author hereby grants to MIT permission toreproduce and to distribute publicly paper and electronic copies of this thesis document in whole or in part.
Signature of Author Ken Huang System Design and Management Program February 2007
Certified By Dr. Ricardo Valerdi Thesis Supervisor Research Associate, Lean Aerospace Initiative
Certified By_ Patrick Hale Director System Design and Management Program
Towards An InformationTechnology Infrastructure Cost Model
By Ken Huang Submitted to the System Design & Management Program On Jan 31st, 2007 in Partial Fulfillment of the Requirements for the Degree of Master of Science in Engineering and Management
Ever since the introduction of the Internet in 1994, one of the defining characteristics of the global economy, particularly in the US, is a dramaticincrease in expenditures on Information Technology. While this trend is expected to continue, a major issue for companies of all sizes is the manner in which precise forecasting of future IT cost may be undertaken. The present thesis investigates the possibility that a set of the essential deterministic cost drivers with varying weighted factors may prove capable of estimating total IT infrastructurecosts. An online questionnaire was developed for this purpose, and was used to survey senior IT leadership teams. The data collected from this survey was then computed with Analytical Hierarchy Process (AHP) to illustrate the relative importance of different cost drivers. The study revealed three primary findings. First, that a set of essential deterministic cost drivers with varying weighted factorscould be used as a general tool for estimating the total cost of IT infrastructure. Second, these different sectors prioritize cost drivers differently from each other. In the Financial Services sector, for instance, the security of the IT network was reported to be of greater importance than the service call response time. In the Technology sector, however, the opposite was true. Third, numerouscorrelations were found to exist within each cost driver category defined. The correlated nature of these cost parameters may mean that a more parsimonious model may be more predictive of total IT infrastructure costs. It is hoped that these findings may be of benefit to a variety of large and small commercial and government entities, which may be able to use the predictive cost drivers to helpeliminate problems related to inaccurate IT cost estimates. It is believed that the cost model proposed may be applicable across a variety of economic sectors. In this thesis, its applicability is demonstrated within the 3
financial services and technology sectors. Future research may be useful in evaluating the model further, by increasing the sample size, and by testing the reliability andvalidity of the cost model within additional economic sectors.
Thesis Supervisor: Title:
Dr. Ricardo Valerdi Research Associate Lean Aerospace Initiative
My ten years of experience in the IT workforce has made me highly sensitive to the increased spending on IT infrastructure in recent years. I am fortunate to have had numerous inspirational colleagues who haveundergone the pain of projecting annual IT budgets with me and who have provided me with useful insight throughout the years. In early October, 2005, Dr. Ricardo Valerdi sent out an email seeking an IT Cost Model researcher from within the MIT community. I was so thrilled and applied for it without any hesitation. I was very fortunate that he has accepted me. Ever since then, I have greatly enjoyed...