STRCUBE - An architecture and Reference Implementation for Streaming Star Schema

Star Schema has been used in data warehouses for creating multi-dimensional models of data for analytics. OLAP operations are performed on the dimensional model for exploratory analytics. Streaming data poses special challenges when managed using traditional database models like vanilla relational data model. The notion of continuous queries and unbounded storage goes against the principles of traditional database architecture. In this paper, we propose STRCUBE, which is an architecture for implementing star schema where the fact table is not a normally stored table but is streaming data. The architecture supports OLAP operations such as slice, dice, rollup, drill down on streaming fact tables. The unique contribution of the work is that the architecture is not tied to any specific streaming platform or bigdata platform. The paper also provides a reference implementation with results from the domain of stock market data.