Toward heterogeneous computing to facilitate sequential OLAP real-time applications

Over the last decade, due to the need of analyzing and studying the logical order that data exhibit in various industries, sequential data storage and processing field has attracted a significant number of researchers. Recently, sequential OLAP has emerged as one of sequential data subfields whereby traditional OLAP - which mainly utilizes a set data-based analysis, do not satisfy the hunger of performing pattern-based operations and time-based analysis. Such analyses can provide an insightful perspective and reveal hidden correlations among events patterns through time. Therefore, extended query languages, new OLAP cube models and optimized algorithms and infrastructures have been introduced. However, the ever grown data size has always been deemed a major hurdle in the way of fully taking advantage of this data. In this context, and based on our proposed optimized heterogeneous Rabin-Karp algorithm earlier, we suggest a high performance sequential pattern detection approach that works in harmony Sequential OLAP processing requirements. The optimized algorithm is dedicated to detect patterns over parallel data streams in Real-Time. The empirical results have shown more than ten times speedup over the multi-core version