Scalable system for e-orders as a service in cloud

Abstract

Cloud market grows each year and companies' dilemma is when and where to migrate their services and applications. Companies can migrate the whole application together with its data, or move only data to some application or service that is already prepared in the cloud. Companies have challenges in either way. For the former, the company should redevelop its application to be applicable to the cloud architecture of the cloud service provider, while it should develop some adapter or make some data transformation for the latter. This paper focuses on the migration of the application along with its data and presents the development of scalable and portable system for e-ordering, which can be offered as a service in the cloud. The system orchestrates the resources dynamically and groups buyers and sellers according to the needs, thus reducing the database size and optimizing the cost and performance. Series of experiments are conducted to prove the scalability of the system hosted in the scalable Eucalyptus cloud framework. The results show a phenomenon, that is, superlinear improvement (greater than the scaled resources) of the performance while scaling the resources. The new architecture achieves speedup of up to 20 while scaling the resources only with factor four. Also, superlinearity appears in the number of handled requests.