Behaviour-Inspired Data Management in the Cloud
Open source cloud computing solutions are still not mature enough to handle data-intensive applications, e.g., scientific simulations. Thus, it is crucial to propose appropriate algorithms and approaches to the data management problem in order to adjust cloud-based infrastructures to scientific community requirements. This paper presents an approach inspired by observation of the cloud user behaviour: the intensity of data access operations, their nature, etc. The authors also describe how the proposed approach influences the architecture of a typical cloud solution and how it can be implemented based on the Eucalyptus system which is a successful open source cloud solution.