On Efficiency and Scalability of Software-Defined Infrastructure for Adaptive Applications
Nasim Beigi-Mohammadi, Hamzeh Khazaei, Mark Shtern, Cornel Barna, Marin Litoiu · 2016
This paper proposes and evaluates a novel analytical performance model to study the efficiency and scalability of software-defined infrastructure (SDI) to host adaptive applications. The SDI allows applications to communicate their adaptation requirements at run-time. Adaptation scenarios require computing and networking resources to be provided to applications in a timely manner to facilitate seamless service delivery. Our analytical model yields the response time of realizing adaptations on the SDI and reveals the scalability limitations. We conduct extensive testbed experiments on a cloud environment to verify the accuracy and fidelity of the model. Cloud service providers can leverage the proposed model to perform capacity planning and bottleneck analysis when they accommodate adaptive applications.