Building Autonomic Systems for IT Service Management

Abstract

A large fraction of total cost of ownership for IT systems comes from service and labor costs (80% as estimated by Gartner Group). However, most prevailing practices in IT services are ad-hoc with low productivity. Although modeling and optimization techniques have been widely used in IT infrastructure management, their adoption in IT service management has limited success due to high uncertainties resulting from complex tasks and large human skill variance. This talk explores the principles and methodologies for building autonomic systems that enable a systematic and feedback-based approach to model and optimize IT service management. Particularly, I will discuss the design and deployment of a staffing (staffing levels, shifts, and skills) self-optimization system that uses the simulation-optimization techniques to enable accurate staffing decision-making, coupled with the feedback-based model validation techniques to address the unique service modeling challenges of employing massive data with large inaccuracies. The developed system has been deployed within the IBM's globally located delivery organizations and resulted in improved resource usage, delivery efficiency, and service quality over a large number of service delivery teams. I will conclude this talk by discussing the future challenges and opportunities of growing autonomic computing capabilities in the IT service management space.