The paper presents the results of investigations concerning an open transport platform that can be used to build grid applications based on object-oriented databases. The described research has been focused on implementation of previously published project assumptions. The presented cases comprise: centralized architecture of a network for a transport platform, utilization of the JXTA technology for building a virtual network, foundations of a communication protocol for the network and an example of its application. An example of a transport platform integrated with an external user application is presented. All the described features are operating within the prototype.
(2009) Kuliberda, Kamil; Kowalski, Tomasz Marek; Wiślicki, Jacek; Adamus, Radosław; Meina, Michał
The problems of integration of distributed resources are currently one of the most substantial issues in the domain of collecting data and retrieving consistent and reliable information - the answer has been included in the following paper. Authors describe aspects of transparent integration of distributed data into an object-oriented data grid with application of the p2p technology and introducing extremely crucial issues of indexing. The presented solution has been implemented and verified in the completely functional prototype. The paper presents basics of application of the p2p architecture and procedures of indexing data originating from remote sources. These procedures accelerate data access by orders of magnitude and data transportation becomes limited to the necessary minimum.
(2009) Kuliberda, Kamil; Kowalski, Tomasz Marek; Wiślicki, Jacek; Adamus, Radosław; Meina, Michał
Authors present the implemented and verified with the fully functional prototype approach to integration of legacy relational data into an object-oriented data grid. The presented process in completely transparent due to application of the virtual repository concept and updatable virtual object-oriented views defined in the stack-based approach (SBA). In the described process much stress has been put on employing native optimizers of relational databases. Hence, transportation and processing of retrieved data within the virtual repository have been limited to the required minimum. Due to the described procedures and architecture, an end user receives a purely object-oriented schema reflecting his/her business requirements and/or access privileges, whose data can be queried with an object-oriented query language (SBQL). Data available in relational systems are retrieved as objects and they can be arbitrarily combined with results coming from other integrated data sources (e.g., object-oriented, relational, XML), so that the user is not aware of their actual origin.
In paper we present an overview of query optimization by indexing for SBQL (Stack-Based Query Language). Developed methods have been implemented and tested in ODRA prototype system. The ODRA index implementation is based on linear hashing and works in a scope of a standalone database. It consists of transparent optimization, automatic index updating and management facilities. The semantic equivalence of proposed query optimization methods in the context of object data model and query language is discussed on several examples.
(2010) Adamus, Radosław; Kowalski, Tomasz Marek; Kuliberda, Kamil; Wiślicki, Jacek; Bleja, Michał
The article is an attempt to describe the current state of development tools whose purpose is building web applications. The article is focused on applications that rely on a relational database as a mechanism for data persistence, covered by an object-oriented application working in a client-server architecture (this should be considered rather as a functional simplification to more complex architectures) available to end users by a web interface. The leitmotif is a thesis that the fundamental element of complexity of such applications is the impedance mismatch between data models used in storage and business logics. The thesis also claims, that a commonly accepted direction of development of such solutions does not lead to any elimination of this inconsistency, but rather tends to hide it inside a mapping layer, which may, in certain cases, lead to a co-unterproductive effect.
(2009) Adamus, Radosław; Kowalski, Tomasz Marek; Kuliberda, Kamil; Draus, Cezary; Wiślicki, Jacek
We describe a generalized approach to the problem of the automatic index updating in response to modification of corresponding data. To enable creation and transparent maintenance of indices supporting keys defined using arbitrary, deterministic and side effects free expressions the authors propose applying a special kind of database triggers. Query language for object-oriented model (classes, inheritance, polymorphism, class methods, etc.) allows easy defining of more complex selection predicates, nevertheless, in order to provide full indexing transparency, index updating requires substantial revising. Inadequate index maintenance can lead to serious errors in query processing. The authors work is based on the Stack-Based Architecture (SBA) and has been implemented in the ODRA (Object Database for Rapid Applications Development) OODBMS prototype.
The paper presents the first stage of accomplishing a DataGrid concept by utilizing an open transport platform based on the peer--to-peer architecture. Implementation details concerning integration of object-oriented databases with the prototype of a transport platform are described. The prototype solution enables one to accomplish a grid network integrating distributed object-oriented data-bases.
