A performance and usability comparison of automated planners for IT change planning

Abstract

Change Management, a core process of the Information Technology Infrastructure Library (ITIL), is concerned with the management of changes to networks and services to satisfy business goals and to minimize costly disruptions on the business. As part of Change Management, IT changes need to be planned for. Despite previous efforts to use planning algorithms to generate IT change plans, it remains unknown as to which automated planning algorithm does best solve the IT change planning problem. To answer this question, we compare four domain independent automated planners in the context of an Infrastructure as a Service deployment case study of a three-tier business application. We focus on two aspects: (1) The scalability of the planners to large Configuration Management Databases (CMDBs) comprising thousands of resources and (2) their usability by a change manager, in particular how easily a change manager can specify the planning domain and rules to guide the search. For the deployment case study we conclude that Hierarchical Task Network algorithms scale to significantly larger CMDBs than all other examined algorithms. Furthermore, we find that their way to formalize search control naturally matches to the domain of IT change planning compared to that of others who require change managers to have a profound knowledge of the planning algorithm or temporal logic.