Predictive Indexing
Joy Arulraj, Ran Xian, Lin Ma, Andrew Pavlo

TL;DR
Predictive indexing uses machine learning to continuously and proactively optimize database index configurations with minimal performance disruption, significantly improving throughput over traditional methods.
Contribution
The paper introduces a novel predictive indexing approach that forecasts index utility and applies small, incremental physical design changes over time.
Findings
Achieves 3.5-5.2x throughput improvement
Works seamlessly with other tuning methods
Uses lightweight hybrid scan operator
Abstract
There has been considerable research on automated index tuning in database management systems (DBMSs). But the majority of these solutions tune the index configuration by retrospectively making computationally expensive physical design changes all at once. Such changes degrade the DBMS's performance during the process, and have reduced utility during subsequent query processing due to the delay between a workload shift and the associated change. A better approach is to generate small changes that tune the physical design over time, forecast the utility of these changes, and apply them ahead of time to maximize their impact. This paper presents predictive indexing that continuously improves a database's physical design using lightweight physical design changes. It uses a machine learning model to forecast the utility of these changes, and continuously refines the index configuration of…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsData Management and Algorithms · Advanced Database Systems and Queries · Data Mining Algorithms and Applications
