Vol. 3 No. 1 2008
Published Issue Open Access

Vol. 3 No. 1 (2008)

Published: June 15, 2008

IADIS International Journal on Computer Science and Information Systems Volume 3, Issue 1, 2008.

Volume 3, Issue 1
Year 2008

Table of Contents

Peer-Reviewed Research
Original Research

Ilona Bluemke, Marcin Warda Institute of Computer Science, Warsaw University of Technol- ogy, Nowowiejska 15/, 00- Warsaw, Poland Researcher

At the Institute of Computer Science Warsaw Un iversity of Technology a module for monitoring ser- vices in Service Oriented Architectures (SOA) was designed and implemented. The module is able to monitor services on the ESB (Enterprise Service Bus) level even, if the services are executed on differ- ent servers. The whole context of the flow is provided to the user. The architecture of this module is briefly presented. This tool was used to measure several parameters in real enterprise integrated architec- ture e.g.: the effectiveness and the usage of resour ces in some flows. These experiments are described and some conclusions are given. The monitoring m odule may be very useful in the maintenance of complex SOA systems. KEWORDS SOA, quality of service, monitoring, platform integration

DOI: 10.33965/ijcsis_2008_v3i1_02

Rafał W. Cegielski, Jarosław A. Chudziak, Joanna Meyer1 1 Prokom Software S.A., Ul. Grójecka, 02- Warsaw, Poland Researcher

In this paper we discuss the role of communication management in the success of a large implementation of IT solutions. It is well known that most of project failures in the recent years are not due to technology shortcomings or lack of technical competency of the project team. It is rather caused by “softer” elements of projects includin g badly planned and executed orga nization transformation, change management process, and not well enough prepared and executed communication management among various participants of the implementation process. In the paper we present some elements of the communication management model: various dimensions of communi cation process, its role and objectives, as well as main participants and several m odern tools and channels which can be used in the process. We give also several examples of their practical usage in real projects.

DOI: 10.33965/ijcsis_2008_v3i1_03

Hyun Seung Lee, Ha Yoon Song, Kyung-Chang Kim Department of Computer

Flash memory can be a viable solution for the fu ture embedded systems. Embedded systems usually carry database as a part of embedded software. It is well known that Flash memo ry is far faster than usual hard disk storages especially for read however it takes much time for Flash memory to delete than to read or to write. For a database system, index tr ees on Flash memory have been widely studied for the better performance. From our previous research es on indexing tree performance on Flash memory, we suggest a delaying technique for index tree ope rations on Flash memory. Among B-tree, R-Tree and MR-tree, we choose MR-tree as it is multimedia data base index. Considering both of the flash memory operation characteristics and MR-tree operation characteristics, we adjust MR-tree parameters for a just fit for Flash memory and applied aggregated-delayed MR-tree operations on Flash memory. We believe we find a reasonable parameter and efficient operation delaying technique for MR-tree on flash memory. The benefit of our researches has been proven by the experimental results based on several database access profiles.

DOI: 10.33965/ijcsis_2008_v3i1_06

IADIS Research Contributor

IONS FOR PROCESS NETWORKS Maciej Koutny*, Giuseppe Pappalardo** and Marta Pietkiewicz-Koutny* (*) School of Computing Science, Newcastle University, Newcastle upon Tyne NE1 7RU, U.K. {maciej.koutny, marta.koutny}@ncl.ac.uk (**) Dipartimento di Matematica e Informatica, Università di Catania, I-95125 Catania, Italy pappalardo@dmi.unict.it ABSTRACT A promising way of dealing with co mplex behaviours of networks of communicating processes is to use abstractions. In our previous work, interface abstraction, modelled through a suitable relation, allowed us to `interpret’ the behaviour of an implementation pro cess as that of a specification process, even in the event that their interfaces differ. The proposed relation is compositional, in the sense that a composition of communicating sub-systems may be implemented by connecting their respective implementations. In our previous work, abstraction has been shown to distribute only over network composition, which restricts its usefulness for compositi onal correctness analysis. In this paper we extend our treatment to other process constructs, known to be useful in the development of complex distributed applications.

DOI: 10.33965/ijcsis_2008_v3i1_07

Michael Sobolewski Computer Science, Texas Tech University, SORCER Research Group, http://sorcer.cs.ttu.edu Researcher

Six generations of RPC systems can be distinguished including Fede rated Method Invocation presented in this paper. Some of them—CORBA, Java RMI, and Web/OGSA servi ces—support distributed ob- jects. However, creating object wrappers implementing remote interfaces doesn’t have a great deal to do with object-oriented distributed programming. Dist ributed objects developed that way are usually ill- structured, difficult to use, with missing core object-o riented traits. A distributed system is not just a collection of distributed objects—it is the network of dynamic objects that come and go. In particular, the object wrapping approach does not help to cope with network-centric messaging, invocation latency, object discovery, dynamic object federations, fault det ection, recovery, partial failure, etc. The Jini™ service architecture does not hide the network; it allo ws the programmer to deal with the network reality to form dynamic service federati ons. However, handling low-level ne tworking details in Jini does not help to cope with complex network programming to dy namically cluster, federate, and utilize efficiently various network services easily. A new network progra mming methodology is presented in this paper. It uses the intuitive metacomputing semantics and the Triple Command design pattern. The pattern defines how service objects communicate by sending one anothe r a form of service messages called exertions that encapsulate data, operations, and control strategy.

DOI: 10.33965/ijcsis_2008_v3i1_08
Case Studies & Applications

Anna Derezińska, Jacek Zawłocki Institute of Computer Science, Warsaw University of Technology, Nowowiejska 15/, 00- Warsaw, Poland Researcher

Model transformation approach allows us to develop automatic and flexible solutions for the software evolution. Application of model transformation concepts was shown on a case study of a generic framework for traceability in object-oriented designs. Traceability analysis in the framework is based on discovering traceability relationships among model elements and identifying dependency areas for given initial elements. The analysis is controlled by a set of rules that should be easily modified ac- cording to the project notation, application domain, etc. Three transformations within the framework were considered: the input transformation of any m odel to the internal format, the traceability analysis generating a dependency area for a given model, a nd the output transformation of the resulting de- pendency area. They can be realized as model-to-m odel transformations with respect to their meta- models and in accordance to the independently speci fied transformation rules. The language and tools of the QVT standard proposed by the OMG were a pplied in the input and output transformations. In the remaining transformation, traceability rules were de fined as automata with transitions labeled with conditions and actions.

DOI: 10.33965/ijcsis_2008_v3i1_04

Piotr Gawkowski, Maciej Ławryńczuk, Piotr Marusak, Janusz Sosnowski, Piotr Tatjewski

The paper studies dependability of software impl ementation of the explicit DMC (Dynamic Matrix Control) Model Predictive Control (MPC) algorithm applied for a rectification column. The process with two inputs and two outputs with strong cross-couplings and significant time delays is studied. The algo- rithm’s control law is calculated off-line. Dependab ility is evaluated experimentally using software implemented fault injection approach. The injected faults influence the quality of rectification process.

DOI: 10.33965/ijcsis_2008_v3i1_05
Editorial

Pedro IsaĂ­as

ions for Process Netw orks, Maciej Koutny, Giuseppe Pappalardo and Marta Pietkiewicz-Koutny - presents a general compositional approach which allows one to deal with abstractions in networks of communicating processes, assuming no restriction about the channels a process can use and about the base processes. • Exertion Oriented Programming, Michael Sobolewski - proposes a new network programming methodology that uses intuitive metacomputing semantics and the Triple Command design pattern. The pattern defines how service objects communicate by sending each other a form of service messages called exertions.

DOI: 10.33965/ijcsis_2008_v3i1_01