'Shared-Nothing' Cloud Data Warehouse Architecture

Abstract

Energy management systems from Romania do not have the capabilities of energy specific management due to lack of technology for real-time monitoring. As was the case in many other countries, the advent of smart metering technology will increase the level of energy data significantly. Therefore, the purpose of this paper is to present solutions that need to be taken to solve problems linked with the increasing amount of data recorded by sensors. For a better demonstration of theoretical elements exposed, we considered a data warehouse specific to utility companies. Section 2 of this article defines the three widely used parallel data warehouse architectures, while in Section 3 we clarify what architecture is suited to develop a data warehouse in the cloud. In the last part we transposed our tables in a "shared-nothing" architecture, trying to analyze queries performance.

Keywords

• Shared-nothing architecture
• Data Warehouse
• Replication
• Distribution
• Cloud