Department: Computer Science and Systems Engineering
Advisor: Marta Lima de Queirós Mattoso, D.Sc.

Nicolaas Ruberg
November/2002

    Cost parameters and database statistics are the basis of query optimization techniques. However, in distributed and heterogeneous database systems, acquiring and treating information to help the optimization process are often tasks of a global query processor, tailoring these functionalities to a specific system architecture. Moreover, this acquisition process involves a large number of parameters and requires customized methods to retrieve data from specific sources.

    DIG (Distributed Information Gatherer) is a cost and statistics provider that, through an independent and flexible service, aims to support global query optimization processing in distributed, heterogeneous database systems over autonomous data sources. The DIG architecture presents two basic components: a module provider of costs and statistics, that does the handling and the publication of the collected data; and module collectors responsible for the data acquisition on diverse sources, which pass the collected data to the according provider. The data acquisition is performed by the DIG collector through the submission of queries or predefined commands (in a specific configuration file) for each statistic or cost parameter being collected. The collected data are published by the DIG provider through a generic catalog of costs and statistics, which supports from semi-structured or non-structured data sources (e.g., text files and web pages) to data sources with sophisticated SGBDs.

    We have developed a DIG prototype and evaluated it with specific wrappers for a query middleware on semi-structured data sources and also for an object-based DBMS.