An Overview of Data-Driven Techniques for IT-Service-Management

High availability of information technology (IT)-applications and -infrastructure components is a significant factor for the success of organizations because more and more business processes depend entirely on IT-services. The result is that the complexity of IT-environments is permanently increasing and therefore cost intensive maintenance processes become more difficult. Several frameworks for IT-service-management like IT Infrastructure Library bundle established best practices to support the task of IT-service operating. This paper provides an overview of data-driven techniques, which can be used in addition to the standards of best practice frameworks to improve reactive and proactive maintenance tasks. We present a selection of different machine learning and statistical methods from a theoretical and practical perspective, which enable detailed insights into the IT-environment status. Furthermore, we present different approaches for the analysis of IT-service tickets, which are the primary communication path between the customer and service provider in case of issues. These methods enhance the resolution-finding process by a higher automatization level or by improvement of the situational understanding in case of root cause analysis and problem determination. Moreover, prediction of the future status allows the initiation of proactive actions to possibly prevent critical situations before their occurrence; existing methods, their potentials, and shortcomings as well are in the center of this paper. We close our overview with a systematic guide for practitioners to select the proper method for their setting, based on a series of questions about the application at hand.