A Model to Compare Cloud and non-Cloud Storage of Big Data

Abstract

When comparing Cloud and non-Cloud Storage it can be difficult to ensure that the comparison is fair. In this paper we examine the process of setting up such a comparison and the metric used. Performance comparisons on Cloud and non Cloud systems, deployed for biomedical scientists, have been conducted to identify improvements of efficiency and performance. Prior to the experiments, network latency, file size and job failures were identified as factors which degrade performance and experiments were conducted to understand their impacts. Organizational Sustainability Modeling (OSM) is used before, during and after the experiments to ensure fair comparisons are achieved. OSM defines the actual and expected execution time, risk control rates and is used to understand key outputs related to both Cloud and non-Cloud experiments. Forty experiments on both Cloud and non Cloud systems were undertaken with two case studies. The first case study was focused on transferring and backing up 10,000 files of 1 GB each and the second case study was focused on transferring and backing up 1,000 files 10 GB each. Results showed that first, the actual and expected execution time on the Cloud was lower than on the non-Cloud system. Second, there was more than 99% consistency between the actual and expected execution time on the Cloud while no comparable consistency was found on the non-Cloud system. Third, the improvement in efficiency was higher on the Cloud than the non-Cloud. OSM is the metric used to analyze the collected data and provided synthesis and insights to the data analysis and visualization of the two case studies.

More information

Identifiers

ORCID for Gary Wills: orcid.org/0000-0001-5771-4088

Catalogue record

Export record

Altmetrics