Applying service-oriented concepts in industrial control and automation systems requires developing new architectural concepts to make it more adaptable to this context. In the scope of service-oriented systems one of the main challenges is defining the right granularity for every service in the system. This paper presents a systematic classification of methods to decompose the services. Identifying and using the correct granularity for each service is expected to lead to higher system flexibility, scalability, and in general to improving service deployment and maintenance. To find the best granularity for a service several approaches have been proposed, some of which are based on modeling and others based on service quality analysis. The approach we propose is based on modeling.