Thesis Open Access
Research infrastructures provide data and other necessary services required by domain scientists for performing advanced research. The performance of the research infrastructure is crucial for the user experience. By analysing the access patterns in the log files data infrastructure operators can improve the quality of the offered end product. This study focuses on the access log files obtained from the file server of the Euro-Argo research infrastructure. Based on the operational history contained in the log files we evaluate how these usage patterns can be used to improve the offered service level to the users of the data infrastructure. We introduce a prediction based model that can forecast the future workload by exploiting the usage patterns extracted from the log files. This model makes it possible to allocate resources in advance leading to a more efficient and optimised data infrastructure improving the service level.