Speaker: Gerald Weber
Title: Efficient Joins between a Stream and Non-Stream Data.
Abstract: Join operators are central to conventional databases as well as to stream databases. In recent years, an interesting type of join has been proposed that has heterogeneous input: one stream and one disk-based relation.
MESHJOIN is a hash join algorithm for this task, I assumes that the stream is non-intermittent.
In the past years we have studied whether the performance of this type of join can be improved and its application domain can be broadened. I will first present HYBRIDJOIN, an algorithm that is suitable for intermittent streams as well. Then I will discuss how we can make use of characteristics of data, in particular non-uniform data.
Short Bio: Gerald Weber is Senior Lecturer in Software Engineering and Computer Science at the University in Auckland, New Zealand. He received his PhD from Freie Universitaet Berlin. His research interests are all aspects of data intensive systems, ranging from the core database technology, over modeling, messaging and concurrency, up to user interfaces. He is the author of over 40 peer reviewed publications and has been active in various chair roles in international conferences. He is Information Director of the Proceedings of the VLDB Endowment.