Please note: This PhD seminar will be given online.
Michael Abebe, PhD candidate
David R. Cheriton School of Computer Science
Supervisor: Professor Khuzaima Daudjee
Hybrid transaction/analytical processing (HTAP) workloads contain a mix of transactional (OLTP) and analytical (OLAP) requests. To meet these demands, distributed HTAP systems typically maintain a complete copy of data in row-oriented storage format that is well-suited for OLTP workloads and a second complete copy in column-oriented storage format optimized for OLAP workloads. Maintaining these data copies consumes significant storage space and system resources. Conversely, storing data in a single format results in performance degradation for one aspect of the workload (OLTP or OLAP).
In this talk, I will present Proteus, a distributed HTAP database system that autonomously selects and adapts its storage layout to optimize for any workload. Proteus generates physical execution plans that utilize storage-aware operators for efficient transaction execution. Using comprehensive HTAP workloads and state-of-the-art comparison systems, I will show how Proteus delivers superior HTAP performance while providing OLTP and OLAP performance on par with designs specialized for processing only one of the two workloads.