Efficient cross-device OLTP/OLAP processing in hybrid CPU/co-processor database systems

点击翻译按钮获取中文摘要

Nowadays, we face two challenges in database systems. First, database systems must combine online transactional processing (OLTP) and online analytical processing (OLAP) workloads to enable real-time business intelligence. Real time business intelligence is needed to improve the quality of created reports and analyses, because up-to-date data is used for analysis instead of historical data, processed in traditional database systems.Second, computer systems become increasingly heterogeneous to provide better hardware performance. Therefore, the architecture of computer systems is changing from single-core CPUs to multi-core CPUs supported by co-processors. Database systems must consider both trends to improve the quality of the systems, to increase the performance, and to ensure that database systems can cope with future requirements (e.g., more complex queries or increased data volume).Unfortunately, current research approaches are restricted to one of these challenges, either combining OLTP/OLAP workloads in traditional CPU-based systems or using co-processors on a single workload type with several restrictions. Therefore, up to now, no holistic approach considering both challenges exists. In this project, we want to face both challenges of database systems to enable efficient processing of mixed OLTP/OLAP workloads in hybrid CPU/co-processor systems, necessary for efficient real-time business intelligence. Hereby, the main challenge is to guarantee the ACID properties, needed for OLTP and mixed OLTP/OLAP workloads, in hybrid systems, while ensuring efficient processing of the mixed workloads.