Exploiting punctuation semantics in data streams

Abstract

Applications that process data streams are becoming common. These applications are often queries over streams, so it seems natural to use a database management system instead of a custom application. However, some traditional relational operators are not conducive to stream processing. We propose embedding punctuations into data streams. A punctuation is a predicate that describes a subset of tuples. It informs a stream processor that no tuples exist after that punctuation that satisfy its predicate.