Skip to Main Content
We consider geographically distributed datacenters forming a collectively managed cloud computing system. Multiple SaaS providers host their SOA-based, context-aware applications in the cloud. Typically, the context-aware applications serve multiple classes of customers (end users) classified on economic considerations, which determine the Quality of Service (QoS) received by each class. This need for differentiated QoS for each customer class is incorporated into a Service Level Agreement (SLA) negotiated between the context-aware application provider and the cloud provider. A QoS metric that has been explored in large distributed applications is the percentile of response times, this metric provides a form of guarantees on the shape of the response time distribution for the customer. Typical SLAs require the response time of a certain percentile of the input requests from particular classes of customers to be less than a specified value, if this value is exceeded, a penalty is charged to the cloud provider. In addition, the applications we consider are data-intensive with strict temporal order constraints that have to be enforced on requests within the same session of a customer. We propose Data-aware Session-grained Allocation with gi-FIFO Scheduling (DSAgS), a novel decentralized request management scheme deployed in each of the geographically distributed datacenters, to globally reduce the penalty charged to the cloud computing system. Our simulation evaluation shows that our dynamic scheme far outperforms commonly deployed management policies (typically employing static or random allocation with First In First Out, Weighted Round Robin or dynamic priority-based scheduling). We further optimize our solution for dynamic, data-intensive context-aware applications, by proposing a "context level" cache replacement policy. Our evaluation shows that, when used in conjunction with DSAgS, the replacement policy decreases the total penalty charged to the cloud.
Date of Conference: 4-9 July 2011