contact    site map  
    polski   english

Databases, Grid Systems and Software Engineering

 

The software engineering and database researches are conducted by prof. Kazimierz Subieta. The activities concerns:
Contemporary computer systems cooperate within a network environment, which, especially in case of Internet, is very heterogeneous. Ones of the most interesting technologies for a data interchange and its integration are peer-to-peer networks. A p2p technology is perceived a cure for the constraints of current TCP/IP based architectures, especially in the context of continuous growth of Internet.
Our researches concerning a peer-to-peer architecture aim to build a foundation for unlimited data processing in a distributed heterogeneous object-oriented database.
In data processing in a distributed heterogeneous environment there is often a need to adapt a local data model to the global canonical one (that is agreed an understood by all data grid participants).
Our team has worked out a generic wrapper for relational databases allowing this kind of data to be accessible through an object-oriented data grid (a virtual repository). The wrapper characterizes with fully transparent data processing that automatically transforms queries targeting a global object-oriented data model into relational ones. Very interesting results have been achieved concerning a query optimization subject. The wrapper is able to generate queries that are able to utilize an optimization engine of a relational data management system.

An inherent and difficult problem in distributed data processing systems is concerned with its performance optimization. In a large database environment, one of the most important optimization techniques is indexing. Such a transparent mechanism is able to significantly decrease an amount of data that needs to be processed to generate a correct response to an user query. Index, as an auxiliary structure, should minimize its load on a database while time consumed by index maintenance should be considerably smaller than a gain from its action. Our researches covers a data indexation that is based on the linear hashing technology.

 

return to the top


Editors:
Radosław Adamus
Tomasz Kowalski

Last modification:
2015-04-15 14:48:11, Radosław Adamus