One Is Not Enough: A Hybrid Approach for IT Change Planning

Abstract

We propose a novel hybrid planning approach for the automated generation of IT change plans. The algorithm addresses an abstraction mismatch between refinement of tasks and reasoning about the lifecycle and state-constraints of domain objects. To the best of our knowledge, it is the first approach to address this abstraction mismatch for IT Change Management and to be based on Artificial Intelligence planning techniques. This has several advantages over previously existing research including increased readability, expressiveness, and maintainability of the descriptions. We developed the foundations of the approach and successfully validated it by applying it to change request planning for TikiWiki, a Content Management System.

Towards solid IT Change Management: Automated detection of conflicting IT change plans

“…We identified a research gap when it comes to detect conflicts among IT change plans in multi-operator Change Management environments and by IT change planning solutions [3], [4], [5], [6], [7], [8]. To prevent the occurrence of conflicting IT changes, threatening to render IT change plans infeasible, we proposed an approach for the automated detection of conflicting IT changes.…”

“…More recent works, such as [4], [5], [7], including our own research [6], [8], propose approaches for IT change planning that reason about the precondition and effects of IT changes within a single plan. For example, the works by Cordeiro et al [5] and Trastour et al [7] focus on the refinement of IT changes, while our hybrid approach [6] addresses refinement and state-based reasoning at the same time. Common to all recent works on IT change planning is, that they are capable of generating a single, logically sound plan as long as the datacenter does not change in between plan generation and execution.…”

“…In addition to that, object-oriented models have been used in previous research by Eilam et al [23] to describe the state of a data center. IT Change plans are generated by different uncoordinated IT operators or by automated IT change planners [3], [6], [8], [4], [5], [7] based on the current state of the OO CMDB (Step 1). Generated plans are stored in a conflicting plan database (C-plan DB), i.e., a database of unscheduled, but not yet verified to be conflict-free, plans in…”

“…To ensure this, ITIL proposes a Change Management process comprising the evaluation, authorization, planning, test, scheduling, implementation, documentation, and review of IT changes [1]. IT Change Planning, an important step of the Change Management process, has been subject to intensive research [3], [4], [5], [6], [7], [8]. This let to IT change planners which can generate a single sound plan that, when executed, achieves a Request for Change.…”

“…The advantage of our approach lies within the tight integration of object-oriented (OO) Configuration Management Databases (CMDBs), frequently used by commercial CMDB systems [10] and suggested by the Common Information Model (CIM) [11] to model the current state of software and hardware in a data center. In addition to that, previously proposed work on IT change planning [3], [4], [5], [6], [7], [8] and scheduling [12], [13] remains unchanged while inter-plan conflicts are prevented. The remainder of this work is organized as follows: In Section II we discuss related work and its shortcomings with respect to the occurrence of conflicting IT changes.…”