Home // DBKDA 2013, The Fifth International Conference on Advances in Databases, Knowledge, and Data Applications // View article

An Approach for Dynamic Materialization of Aggregates in Mixed Application Workloads

Authors:
Stephan Müller
Carsten Meyer
Hasso Plattner

Keywords: caching, data aggregation, dynamic view mate- rialization, mixed-workload, OLAP, OLTP

Abstract:
Aggregate queries are one of the most resource intensive operations for databases systems. Despite scanning and processing large data sets, they only return relatively small outputs, making them predestined for reuse in future queries. The challenge is to manage the infinite number of possible data areas that can be selected for caching, identify the relevant ones and materialize the corresponding aggregates on a reusable level of granularity. While analytical database systems often save materialized data in cubes of aggregated chunks this is not feasible in systems with transactional insert and change processes. We present the concept dynamic materialized aggregate views (DMAV) that materializes aggregates based on mixed application workloads. Using control tables, aggregate query structures as well as selected data areas are captured and kept track of. We evaluate data access characteristics and only materialize aggregates that are read often but at the same time have only few updates resulting in low aggregates maintenance costs.

Pages: 77 to 86

Copyright: Copyright (c) IARIA, 2013

Publication date: January 27, 2013

Published in: conference

ISSN: 2308-4332

ISBN: 978-1-61208-247-9

Location: Seville, Spain

Dates: from January 27, 2013 to February 1, 2013