Towards an Adaptive QoS of Cloud-based Web Services

Service oriented applications define the Quality of Services (QoSs) in terms of a Service Level Agreement (SLA) which guaranties the non-functional attributes for the users. However, web services usually have unpredictable load conditions and fluctuating demands which makes it difficult to maintain the predefined SLA. For example, extreme access of service-based application, such as e-commerce or news web services connected to a database. It is challenging to provide the expected QoS attributes, such as the response time. Cloud computing has gained a huge interest in academia and major IT organizations. The ability to create several virtual machines on each physical machine (PM), and to resize the virtual machine resource configuration up and down as needed, gives the cloud the ability to guarantees the SLA for the service-based applications. This paper presents a framework for adaptive QoSs of web services in a cloud environment. In case of predictions of the SLA violations, a dynamic resource scaling is managed based on two level. The first level, resources such as CPUs, memory of the virtual machine, is scaled up. The second level, a new virtual machine is instantiated on another physical machine to process the increasing requests using replication. The framework monitors the requests arrival rate and the response times of the web service, and uses a time series to predict the future arrival rates and response times. The process of scaling the resources and virtual machines up and down is based a queuing network model. An experimental analysis is conducted to explore the feasibility of the proposed framework on a private cloud, using Eucalyptus, and a synthetic workload.