Descrizione

Motivation: Temporal data is ubiquitous, and its importance has been witnessed by the recent introduction of temporal features in the SQL standard and commercial database management systems (DBMSs). Despite such efforts, current DBMSs offer little support for query formulation and processing. This makes application code and SQL queries complex and inefficient. Temporal alignment is the first and only framework that provides comprehensive temporal query support over interval-timestamped data and allows a tight integration into existing DBMSs (http://tpg.inf.unibz.it). While the approach minimizes the changes of the host DBMS and leverages its query optimization techniques, important aspects of the efficiency of the query processing remain as unaddressed open problems.

Objectives: This project aims at boosting the efficiency and scalability of query processing in the temporal alignment framework by tackling the following challenges: 

1. The current approach works with two primitives to transform temporal queries into non-temporal queries, which might produce large and redundant intermediate relations. In order to avoid this, we will explore new alignment primitives that are customized for specific temporal operators. 
2. Since current cost estimates for intermediate relations are very conservative, we will develop more realistic cost estimates to feed into the query optimizer. 
3. We will invent new algebraic equivalence rules for query optimization that exploit specific properties of temporal operators. If successful, the project will provide the foundation for the first industrial-strength, open-source RDBMS with comprehensive temporal query support.

Research Methods: The research methods include basic research, technology development and transfer, and empirical evaluation. We will study fundamental research problems in temporal databases and will develop solutions that advance the state of the art. The scientific results will be published in top-level database outlets. A strong emphasis will be on technology development: all results will be implemented in the kernel of the open-source PostgreSQL system and thoroughly evaluated. In terms of technology transfer and dissemination, we plan to submit our results as patch to the PostgreSQL community as well as to run a case study with the local industry.

Project Members: The project is a collaboration between three research partners:

• Free University of Bozen-Bolzano: Johann Gamper (PI), Anton Dignos (Co-PI)
• University of Zurich: Michael H. Böhlen
• Aalborg University: Christian S. Jensen

All three partners have collaborated extensively on temporal database research. Recently, they proposed the temporal alignment framework as the first systematic solution to support temporal query processing, which forms the starting point for this project.

Partner

Lead Partner Libera Università di Bolzano, Facoltà di Ingegneria