No description available.

Data warehousing and On-line Analytical Processing (OLAP) are powerful tools for processing and analyzing business and analytical data. It is estimated that 80% of the data stored in data warehouses have some spatial components. It is our belief that there is a need for powerful OLAP tools that are capable of processing and analyzing spatial data. This thesis explores the design and implementation of Spatial OLAP (SOLAP) systems and describes approaches to support the characteristic features of OLAP while seamlessly integrating spatial data into the analysis process. In particular, we analyze the evaluation of OLAP queries in the presence of asymmetric, multiple-alternative, generalized, and non-strict spatial dimension hierarchies. We introduce a new pipeline-based query evaluation model that is comprehensive and powerful in that it provides a uniform approach to the expression of spatial OLAP queries that address all major dimension hierarchy types while permitting a uniform treatment of both spatial and non-spatial data. A reference implementation called "LISA" validates the objectives of our model and demonstrates favorable scalability and performance on modern multi-processor and multi-core hardware platforms. We also describe a new "geoCUBE" index, to address the fundamental problem of how to represent, index and efficiently query data that is defined by a mix of spatial and categorical attribute values. The geoCUBE index extends existing methods for indexing OLAP data to spatial data types. The effectiveness of the geoCUBE data structure is confirmed through evaluation. Lastly, we propose algorithms that facilitate OLAP-like analysis of moving object data. We introduce a new class of GROUP BY operators specifically targeted to the OLAP analysis of trajectories and to answering aggregate queries with respect to the spatio-temporal movement of a set of objects. Through an experimental evaluation we show our operators can be used to reliably identify groups of related trajectories when applied to synthetic and real world moving object data.

No description available.

No description available.

Preci* is a generalised distributed database management system, capable of supporting heterogeneous, pre-existing databases as nodes. The system is fully decentralised, supporting both retrieval and update of the data. Varying degrees of location transparency can be provided, according to user requirements. The work presented here is concerned with data integration and query decomposition. An extended relational algebra (PAL) is developed, which serves both as a query language and as a mapping language for data integration. The suitability of PAL for data integration is demonstrated by a number of examples, and by comparison with existing proposals. A major attraction of PAL is that it can also be used as a query language, thereby making query decomposition much easier. The relational algebraic approach is shown to be particularly appropriate for query decomposition, since queries can be easily parsed and represented in tree form. Such parse trees are readily transformed to yield equivalent expressions which will execute more efficiently. An algorithm is given for decomposing global PAL queries into nodal subqueries, and for coordinating their execution. The general problem of allocating subqueries to execution nodes is not tackled, though it is shown that the algorithm will do this allocation under specific implementation conditions. A prototype of Preci* has been implemented in 'C'.

The upsurge in the number of casual users and the general acceptance of computer technology has shown that the principal driving force in software engineering is shifting from functionality to usability. It has also become evident that the desktop metaphor and its whole related genre do not provide the modern interface designer with the expressive power that is needed. Nowhere is a new approach more needed than in interfaces for databases. Previous studies in interface design for database management systems have attempted to use solely the desktop metaphor. We have used three dimensional graphical techniques to construct an interface, called <I>Amaze</I> for our object oriented database <I>P/FDM</I>. Interactive animated 3D graphics have been embedded inside a standard menu driven framework. Using 3D graphics new metaphors have been developed to aid the user interaction. Our development has taken a modular approach, which allowed us to develop a number of different visualisations for query construction, the structure of the database and result representation. It is possible to view the data using a number of multimodal displays (a number of customised multimodal displays have been built). <I>Amaze</I> has been used on a variety of different data sets (including a protein structure database, a personnel database and an antibody database). These databases differ greatly in size and complexity of their semantics. The work discussed in this thesis suggests an alternative approach to user interface design for database systems, it introduces the idea of <I>Database Visualisation</I> and suggests novel mechanisms for computer interaction using 3D graphics.

Almost all of the database products developed over the past few years are based on what is called the relational approach.The purpose of this thesis is to characterize a relational data base management system, we do this by studying the relational model in some depth.The relational model is not static, rather it has been evolving over time. We trace the evolution of the relational model. We will also consider the ramifications of the relational model for modern database systems. / Department of Computer Science

Various High Availability DataBase systems (HADB) are used to provide high availability. Pairing an active database system with a standby system is one commonly used HADB techniques. The active system serves read/write workloads. One or more standby systems replicate the active and serve read-only workloads. Though widely used, this technique has some significant drawbacks: The active system becomes the bottleneck under heavy write workloads. Replicating changes synchronously from the active to the standbys further reduces the performance of the active system. Asynchronous replication, however, risk the loss of updates during failover. The shared-nothing architecture of active-standby systems is unnecessarily complex and cost inefficient. In this thesis we present SHADOW systems, a new technique for database high availability. In a SHADOW system, the responsibility for database replication is pushed from the database systems into a shared, reliable, storage system. The active and standby systems share access to a single logical copy of the database, which resides in shared storage. SHADOW introduces write offloading, which frees the active system from the need to update the persistent database, placing that responsibility on the underutilized standby system instead. By exploiting shared storage, SHADOW systems avoid the overhead of database-managed synchronized replication, while ensuring that no updates will be lost during a failover. We have implemented a SHADOW system using PostgreSQL, and we present the results of a performance evaluation that shows that the SHADOW system can outperform both traditional synchronous replication and standalone PostgreSQL systems.

No description available.

No description available.