Tesis:

A PROPOSAL FOR MODEL-BASED AUTOMATION OF ENTERPRISE SERVICE CHANGE MANAGEMENT PROCESSES.


  • Autor: CUADRADO LATASA, Félix Aurelio

  • Título: A PROPOSAL FOR MODEL-BASED AUTOMATION OF ENTERPRISE SERVICE CHANGE MANAGEMENT PROCESSES.

  • Fecha: 2009

  • Materia: Sin materia definida

  • Escuela: E.T.S. DE INGENIEROS DE TELECOMUNICACION

  • Departamentos: INGENIERIA DE SISTEMAS TELEMATICOS

  • Acceso electrónico: http://oa.upm.es/4294/

  • Director/a 1º: DUEÑAS LOPEZ, Juan Carlos

  • Resumen: In a globalized world enterprises have to face greatly increased competition, demanding agility to release new products and update to customer demands. From a technological perspective, these factors have lead to the adoption of the service oriented paradigm, which must be supported by a robust IT infrastructure. One of the main competitive factors is the quality of service provided, ensuring the elements of the services portfolio have high availability, and unnoticeable response times. These non functional requirements are partially supported by the execution infrastructure, composed by multiple, heterogeneous servers with specialized roles, distributed over a network. However, the combination of these factors greatly complicate technical management processes of the infrastructure such as diagnosing the environment status, planning the required changes or applying corrections to improve its performance. Frequently those tasks are manually executed by an IT administrator, but this approach is very costly and hampers the desired agility. An increased degree of automation in service change management operations is a must for obtaining the potential advantages of the service oriented approach. This dissertation proposes an enterprise service management architecture with automated operation capabilities. One of the cornerstones of this proposal is an information model of all the relevant management information. The proposed model builds upon the common ground of the main information model standards to characterize both the logical artifacts, originated from the service development process, and the managed runtime elements, ranging from hardware nodes to the provisioned services. The model not only allows to represent different environment configurations but also provides well defined expressions for validating the correctness of any system state, and automatically obtain the required configuration values for some of the managed resources. In addition to the information model, the business objectives, desired functionality, and changes to the domain have been defined using the same concepts. This way, the effect of external changes to the environment configuration, as well as its impact on the stability and functionality of the environment can be automatically analyzed. After defining all the relevant management information through a cohesive model that covers both technical and business aspects, this dissertation proposes an algorithm for automating the execution of service configuration change management activities, based on pseudo-boolean SAT techniques. The proposed algorithm analyzes the current state of a managed domain and, in case the situation is not stable or desirable obtains the set of required changes to restore the system to its intended functionality. Instead of defining separate processes for installation, reconfiguration, or removal of selected elements, the same reasoning steps produce a change plan with the necessary operations. Finally, after taking into account the requirements of enterprise applications, an architecture for a service change management system has been proposed, based on the described models and reasoning techniques. A prototype of the proposed architecture and algorithms has been developed and validated through a set of case studies taken from the context of a real banking organization. The results of the validation show how different situations such as initial provisioning or reaction to hardware malfunctions are correctly addressed by the architecture, as well as how the proposal scales with increasingly larger environments and defined services.