Abstract

The growing awareness that effective Information Systems (IS), which contribute to sustainable business processes, secure a long-lasting competitive advantage has increasingly focused corporate transformation efforts on the efficient usage of Information Technology (IT). In this context, we provide a new perspective on the management of enterprise information systems and introduce a novel framework that harmonizes economic and operational goals. Concretely, we target elastic n-tier applications with dynamic on-demand cloud resource provisioning. We design and implement a novel integrated management model for information systems that induces economic influence factors into the operation strategy to adapt the performance goals of an enterprise information system dynamically (i.e., online at runtime). Our framework forecasts future user behavior based on historic data, analyzes the impact of workload on system performance based on a non-linear performance model, analyzes the economic impact of different provisioning strategies, and derives an optimal operation strategy. The evaluation of our prototype, based on a real production system workload trace, is carried out in a custom test infrastructure (i.e., cloud testbed, n-tier benchmark application, distributed monitors, and control framework), which allows us to evaluate our approach in depth, in terms of efficiency along the entire SLA lifetime. Based on our thorough evaluation, we are able to make concise recommendations on how to use our framework effectively in further research and practice.

Share

COinS
 

Dynamic Service Level Agreement Management for Efficient Operation of Elastic Information Systems

The growing awareness that effective Information Systems (IS), which contribute to sustainable business processes, secure a long-lasting competitive advantage has increasingly focused corporate transformation efforts on the efficient usage of Information Technology (IT). In this context, we provide a new perspective on the management of enterprise information systems and introduce a novel framework that harmonizes economic and operational goals. Concretely, we target elastic n-tier applications with dynamic on-demand cloud resource provisioning. We design and implement a novel integrated management model for information systems that induces economic influence factors into the operation strategy to adapt the performance goals of an enterprise information system dynamically (i.e., online at runtime). Our framework forecasts future user behavior based on historic data, analyzes the impact of workload on system performance based on a non-linear performance model, analyzes the economic impact of different provisioning strategies, and derives an optimal operation strategy. The evaluation of our prototype, based on a real production system workload trace, is carried out in a custom test infrastructure (i.e., cloud testbed, n-tier benchmark application, distributed monitors, and control framework), which allows us to evaluate our approach in depth, in terms of efficiency along the entire SLA lifetime. Based on our thorough evaluation, we are able to make concise recommendations on how to use our framework effectively in further research and practice.