Adaptive Indexing in Modern Databases Tutorial at EDBT/ICDT 2012
PlanetData partner, Centrum Wiskunde & Informatica (CWI), The Netherlands, together with the HP Labs, Palo Alto, delivered a tutorial during the EDBT/ICDT 2012 Joint Conference held in Berlin, Germany on March 26-30, 2012. Their tutorial entitled "Adaptive Indexing in Modern Databases".
Abstract of this tutorial:
Physical design represents one of the hardest problems for database management systems. Without proper tuning, systems cannot achieve good performance. Traditional indexing creates indexes a priori assuming good workload knowledge and enough idle time. More recent approaches monitor the workload trends and create or drop indexes online, i.e., during query processing. Adaptive indexing takes another step towards completely automating the tuning process of a database system, by enabling incremental and partial online indexing. The main idea is that physical design changes continuously, adaptively, partially, incrementally and on demand while processing queries as part of the execution operators. As such it brings a plethora of opportunities for rethinking and improving every single corner of database system design.
They analyzed the research space between traditional indexing and adaptive indexing through several state of the art indexing techniques, e.g., what-if analysis and soft indexes. We will discuss in detail adaptive indexing techniques such as database cracking, adaptive merging, sideways cracking and various hybrids that try to balance the online tuning overhead with the convergence speed to optimal performance. In addition, we will discuss how various aspects of modern techniques for database architectures, such as vectorization, bulk processing, column-store execution and storage affect adaptive indexing. Finally, we will discuss several open research topics towards fully automomous database kernels.