Proceedings of the International Multiconference on ISSN 1896-7094
Computer Science and Information Technology, pp. 243 – 252 © 2006 PIPS
Spatial Telemetric Data Warehouse and Software Agents as Environment
to Distributed Execute SQL Queries
Marcin Gorawski
1
, Ewa Płuciennik
1
1
Silesian Technical University, Institute of Computer Science,
44-100 Gliwice, Poland
{Marcin.Gorawski, Ewa.Pluciennik}@polsl.pl
Abstract. The article presents environment for distributed SQL query execution in a telemetric data
warehouse. A parallel spatial data warehouse stores information remotely read from meters grouped in nodes
according to their geographical positions. This data warehouse constitutes distributed data structure. Software
agents environment enables querying this structure as local one so distribution is transparent to the user.
Query evaluation consists of analysis, decomposition, local execution and results merging. There is no need
to transfer data between nodes. Authors present operators for executing query modification and sub-results
merging and also test outcomes of SQL query realization in different agent environment configurations.
1 Introduction
For many years in database realm there exists tendency to create systems designated to store and process more
and more huge data volumes measured even in petabytes. It results from rapidly growing amount of information
which human being is unable to analyze. These data are often stored in data warehouses and used for On-line
Analytical Processing (OLAP) or Decision Support Systems (DSS). There are several ways to improve
efficiency (query response time) of such systems: appropriate indexing, materialized views, data partitioning
and parallel processing [1]. The last technique is very interesting because it is natural for organization which
consists of a few or more autonomous geographically distributed departments. In this case each department has
its own local independent system where data is stored for example in autonomous data warehouses. Still there
exists some kind of a central system (for example DSS) which should have access to the data from all
subsystems. Moreover, although hardware capabilities in data transfer, storage and analysis are still rising, it
does not make a guarantee for effective data processing which can be provided by the system enabling parallel
data processing and tasks execution [2]. Parallel data processing can be applied in local multi-processor
environment as well as geographically distributed autonomous machines.
Work related to the query execution problem in distributed Database Management Systems (DBMS)
concentrates on query evaluation plan (QEP) analysis and optimization [3] and also minimization of total
amount of data transferred between nodes [4]. In case of distributed data warehouse very important issue is a
way of distribution. One of such methods has been proposed for star schema in [1]. It is data warehouse
stripping in which dimension tables are replicated and a fact table is uniformly distributed among all computers
in the distributed system. The way of query evaluation in this kind of distributed data warehouse has been
presented in [1, 5, 6].
In this paper we present solution which is a combination of parallel, distributed data warehouse and software
agents environment. Query evaluation is very similar to the one proposed in publications mentioned above. We
243