Workload aware hybrid partitioning

Padiya, Trupti; Kanwar, Jai Jai; Bhise, Minal

Abstract

Real life databases exhibit highly skewed access patterns. These skewed access patterns can be exploited to partition the data considering the query workload. The presented work proposes Workload Aware Hybrid Partitioning (WAHP). WAHP identifies clusters of attributes which are queried together. It identifies workload aware clusters for the actual query workload using a hybrid combination of horizontal and vertical partitioning. The paper demonstrates WAHP experiment using TPC-C benchmark, where 9% of the actual TPC-C data in workload aware clusters, is able to answer 73% of hottest query-workload with an average execution time gain of 37% against original database.

• Presentation Conference Type: Conference Paper (published)
• Conference Name: ACM COMPUTE '16: Ninth Annual ACM India Conference
• Start Date: Oct 21, 2016
• End Date: Oct 23, 2016
• Online Publication Date: Oct 21, 2016
• Publication Date: Oct 21, 2016
• Deposit Date: Sep 13, 2023
• Pages: 51-58
• Book Title: COMPUTE '16: Proceedings of the 9th Annual ACM India Conference
• Public URL: https://uwe-repository.worktribe.com/output/10937862