Modeling Elasticity Trade-Offs in Adaptive Mixed Systems

摘要

In the past decade, elastic computing has emerged as a popular solution approach for on-demand computing resources' provisioning. Today, such dynamic resource provisioning is used not only in machine-based computing systems but also in mixed systems containing machine- and human-based computing elements. Many efforts have been undertaken to achieve resource elasticity. However, to provide a truly elastic computation, system designers also need to foresee trade-offs between cost, quality, and resources. Unfortunately, the lack of modeling tools leads to difficulty in designing on-demand provisioning strategy in mixed systems. In this paper we present a modeling tool named Elasticity Profile (EP) that specifies constructs for modeling the elastic behavior of mixed systems with respect to trade-offs between cost, quality, and resources. We also present a conceptual framework where our EP can be deployed, utilized, and bound to runtime systems. Furthermore, we demonstrate the suitability of EP model using several use cases.