---

TITLE: Quantifying the Complexity of IT System Management

ABSTRACT:
Today, the cost of managing enterprise information technology (IT) systems dwarfs the cost of the systems themselves. Much of this cost can be directly traced to the complexity of the procedures and processes required to keep IT systems up and running. This complexity is a major impediment to the adoption of new IT technology and greatly increases the cost of IT services. It is our belief that this complexity will only be reduced when we can quantitatively measure it and its impact on overall IT costs: just as quantitative performance analysis has been the foundation for the vast improvements in CPU and database performance over the past decades, quantitative approaches to IT management complexity will drive great decreases in total IT cost over the next decade and beyond.

At IBM Research, we are tackling the problem of defining quantitative approaches to IT complexity from two directions. First, we are studying the complexity of low-level configuration tasks—installing, configuring, and maintaining software, hardware, and application solutions—and are developing models and quantitative benchmarks to measure the complexity of these tasks. Concurrently, we are addressing complexity at the upper layers of the management stack, investigating the sources of complexity and cost in high-level, enterprise-wide IT management processes (such as change and configuration management, software distribution, and problem management processes). In the latter case, we are developing analysis techniques and tools that will help the designers of enterprise-level IT processes identify and quantify complexity, and potentially guide complexity reduction through intelligent application of automation technology.

In this talk, I will describe these efforts in detail and present some initial results showing the potential benefits of taking a quantitative approach to IT management complexity. Along the way, I will draw on our experiences working with IBM’s Global Services division to provide additional insight into the complexity challenges facing the industry today in the enterprise IT management space.

BIO:
Aaron B. Brown is a Research Staff Member in the Adaptive Systems department at IBM’s T.J. Watson Research Center in Hawthorne, NY. His research interests include understanding the role and impact of human system managers in large-scale IT infrastructures, quantifying and reducing IT management complexity, improving the dependability of business systems, and benchmarking non-traditional aspects of IT systems. He is also one of the architects of IBM’s Autonomic Computing effort. Before joining IBM, he co-founded the Recovery-Oriented Computing (ROC) project at the University of California, Berkeley, with his PhD advisor (and now ACM President) David A. Patterson.

--