Monetary Cost Optimizations for Hosting Workflow-as-a-Service in IaaS Clouds

摘要

Recently, we have witnessed workflows from science and other data-intensiveapplications emerging on Infrastructure-asa-Service (IaaS) clouds, and manyworkflow service providers offering workflow as a service (WaaS). The majorconcern of WaaS providers is to minimize the monetary cost of executingworkflows in the IaaS cloud. While there have been previous studies on thisconcern, most of them assume static task execution time and static pricingscheme, and have the QoS notion of satisfying a deterministic deadline.However, cloud environment is dynamic, with performance dynamics caused by theinterference from concurrent executions and price dynamics like spot pricesoffered by Amazon EC2. Therefore, we argue that WaaS providers should have thenotion of offering probabilistic performance guarantees for individualworkflows on IaaS clouds. We develop a probabilistic scheduling frameworkcalled Dyna to minimize the monetary cost while offering probabilistic deadlineguarantees. The framework includes an A*-based instance configuration methodfor performance dynamics, and a hybrid instance configuration refinement forutilizing spot instances. Experimental results with three real-world scientificworkflow applications on Amazon EC2 demonstrate (1) the accuracy of ourframework on satisfying the probabilistic deadline guarantees required by theusers; (2) the effectiveness of our framework on reducing monetary cost incomparison with the existing approaches.