The expansion of ubiquitous virtual and physical sensors, leading up to the Internet of Things, has accelerated the rate and quantity of data being generated continuously. Application QoS is also impacted by variability of resource performance exhibited through clouds and hence necessitates autonomic methods of provisioning elastic resources to support such applications on cloud infrastructure. The proposed work is to develop the concept of “dynamic dataflows” which utilize alternate tasks as additional control over the dataflow's cost and QoS. The application model is developed for dynamic dataflows as well as the infrastructure model for representation of IaaS cloud characteristics and an optimization problem is proposed for resource provisioning that balances the resource cost, improves application throughput and improves domain value based on user-defined constraints that are presented through a Particle Swarm Optimization (PSO) based heuristic for deployment and runtime adaptation of continuous dataflows to solve the optimization problem. Also the proposed efficient greedy heuristics can provide optimal solution over efficiency, which is critical for low latency streaming applications. Elasticity is to mitigate the effect on variability, both in input data rates and cloud resource performance, to meet the QoS of fast data applications.
展开▼
