Query Results over Ongoing Databases that Remain Valid as Time Passes By (Extended Version)

TL;DR

This paper introduces a novel approach for handling ongoing time points in databases by evaluating predicates at all reference times, enabling query results that remain valid as time passes, unlike traditional instantiation methods.

Contribution

It proposes a method to process ongoing time points without instantiation, maintaining valid query results over time, and implements this in PostgreSQL.

Findings

Efficient implementation of ongoing data types in PostgreSQL.

Query results remain valid over time without re-evaluation.

Improved handling of ongoing time points in database systems.

Abstract

Ongoing time point now is used to state that a tuple is valid from the start point onward. For database systems ongoing time points have far-reaching implications since they change continuously as time passes by. State-of-the-art approaches deal with ongoing time points by instantiating them to the reference time. The instantiation yields query results that are only valid at the chosen time and get invalidated as time passes by. We propose a solution that keeps ongoing time points uninstantiated during query processing. We do so by evaluating predicates and functions at all possible reference times. This renders query results independent of a specific reference time and yields results that remain valid as time passes by. As query results, we propose ongoing relations that include a reference time attribute. The value of the reference time attribute is restricted by predicates and functions on ongoing attributes. We describe and evaluate an efficient implementation of ongoing data types and operations in PostgreSQL.