A CSP Approach to IT Service Management

An important task of IT service management with respect to the service life cycle is to plan the provision of services in a management environment. The goal is to provide services to customers with a high quality of service by optimizing the use of resources necessary for the realization of the services. IT service managers are constantly confronted with ”look-ahead” problems, like how many new customers can be supported with existing resources. Such problems can be effectively and efficiently solved with a constraint-based approach, which has four key characteristics. (i) It can describe the management environment (i.e., build a model) in an incremental way. (ii) It can deal with incomplete data. (iii) It can cope with changes in highly dynamic environments. (iv) It can find optimal solutions based upon the established model. Problems in planning IT services can be regarded as Constraint Satisfaction Problems (CSPs). A CSP is defined by a set of variables, a set of values for variables, and a set of constraints over variables. A solution to the CSP is an assignment of variables (i.e., a value for each variable) such that all constraints are satisfied. From the modelling viewpoint, the CSP model defines all solutions to the problem whereas incomplete problem data represent a missed constraint for which a larger set of solutions is obtained, and changes in the management environment define modified constraints. The quality of solutions depends on how good the CSP model represents the problem. From the viewpoint of solving, backtracking is a simple algorithm that solves CSP problems. However, this algorithm suffers from a lack of efficiency. The aim is to use other algorithms to solve the problem faster. These can be achieved by (i) developing a dedicated algorithm for a given problem or (ii) decomposing the problem into subproblems and solving them with well-known algorithms. We sketch the proposed approach for the IP service. One of the important questions herewith is to recognize how many new customers (i.e., customer service requests) can be handled with the existing network infrastructure of a provider. There