Abstract

The need for highly available services is ever increasing in various domains ranging from mission critical systems to transaction based ones such as online banking. The Service Availability Forum (SAForum) has defined a set of services and related Application Programming Interface (API) specifications to address the growing need of commercial-off-the-shelf high availability solutions. Among these services, the Availability Management Framework (AMF) is the service responsible for managing the high availability of the application services. To achieve this task, an AMF implementation requires a specific logical view of the organization of the application’s services and components, known as an AMF configuration. Any AMF configuration must be compliant to the concepts and constraints defined in the AMF specifications. The process of defining AMF configurations is error prone and requires extensive domain knowledge. Another major issue is being able to analyze the designed AMF configuration to quantify the anticipated service availability. This requires a different set of modeling and analysis skills that system integrators might not necessarily possess. In this dissertation we propose the automation of this process. The premise is to define a generation method within which we embed the domain knowledge and the domain constraints, and by that generating AMF configurations that are valid by construction. We also define an approach for the service availability analysis of AMF configurations. Our method is based on generating an analysis stochastic model that captures the middleware behavior and the application configuration. This model is thereafter solved to quantify the service availability.