Zugl.: Bamberg, Univ., Diss., 2009. / Zsfassung in dt. Sprache.

Enterprise JavaBeans (EJB) ermöglichen die Erstellung von leistungsfähigen, mehrschichtigen Client/Server-Anwendungen auf Basis der Programmiersprache Java. Eine grundlegende und wichtige Entscheidung, die häufig unterschätzt wird, ist die Art und Weise, wie Daten zwischen komplexen Java-Clients (Fat-Clients) und Servern transportiert werden sollen. Während der Implementierung kann ein fehlendes Konzept zur Übertragung von Daten dazu führen, daß Entwickler unterschiedliche Verfahren wählen, die verschiedenen Anforderungen an die Anwendung, wie z.B. ein möglichst gutes Leistungsverhalten, widersprechen und evtl. unter hohem Aufwand rückgängig zu machen sind. Verschiedene Übertragungsverfahren erschweren außerdem die Erweiterung und Wartung des Systems, da zu analysieren ist, wie in verschiedenen Fällen die Datenübertragung gelöst wird. In diesem Beitrag werden deshalb neue Vorgehensweisen vorgestellt, um Daten zwischen einer objektorientierten Applikationsschicht, die sich auf dem Server befindet, und den Clients zu übertragen. Aktive Daten-Container (ADCs) stellen einen einheitlichen Transportmechanismus bereit, der zusätzliche Funktionalität besitzt, um die Datenübertragung zu optimieren. Dazu gehört z.B. der automatische Datenaustausch mit Geschäftsobjekten, um das manuelle Beschreiben und Auslesen des Daten-Containers durch den Anwendungsentwickler einzusparen. Zusätzlich wird die Optimierung des Datenübertragungsvorgangs im Sinne des Leistungsverhaltens angestrebt. ADCs sind sehr flexibel und können in der Anwendungsentwicklung als universelles und zentral zur Verfügung gestelltes Datenübertragungskonzept dienen. Eine Untersuchung des Leistungsverhaltens zeigt, daß mit den vorliegenden Konzepten der Transaktionsdurchsatz hinsichtlich der Kommunikation bis zum Faktor 8 gesteigert werden kann. Im Rahmen eines Industrieprojekts konnte mit Hilfe der Konzepte Entwicklungsaufwand im Umfang von ca. 18% der erforderlichen Codezeilen einer komplexen Anwendung eingespart werden.

Architectural design decisions are known to be crucial for the success of a system in development. An early design decision will most likely be expensive to change at a later stage of the development if the software engineers needs to get back to the drawing board. If a quality attribute has been neglected during architectural design it may ruin the entire project. This work describes a way of increasing the availability of a software service, that have already been put into use, by adding a middle layer of replication logics represented by third party application servers and replicated Enterprise JavaBeans. The approach called Three-tier replication is assessed and compared to the origin architectural design by using a qualitative scenario based assessment. In addition, we have also implemented the new architectural design to be able to validate the assessment results. The validation is done by testing the scenarios on the two architectures. The software service used is an industrial system for sending text messages from computers to mobile phones. / Tidiga designbeslut är extremt viktiga för utgången av ett utvecklingsprojekt. Felaktiga beslut kommer att kosta företaget eller organisationen mycket tid och pengar om det visar sig att de måste omarbetas. Detta arbetet beskriver ett sätt att öka åtkomsten på en tjänst genom att lägga till ett mellan-lager som sköter replikering av tjänstenoden. Relplikeringslogiken sköts av applikationsservrar och Enterprise Javabeans. Vi har använt ett befintligt industriellt system för att utveckla och jämföra metoden. Tjänsten ifråga är en SMS-tjänst för kommunikation mellan dator och mobil-telefon och vise versa.

<p>The aim of this work is to design an experiment, where client-server systems can be compared. They belong to the group of systems that are called component systems. Client-server systems are difficult to compare, because they are complex. The client-server systems are documented in different ways. Notations in the implementation of them are in different ways. There is a large difference in the communication between the client and server. The architectures between the client-server systems differ also, but they are not totally different. Therefore it is possible to construct an experiment for comparing them. Client-server systems that will be compared are Java RMI, Web Services, CORBA, and Enterprise JavaBeans. We are going to use Java as the programming language. Some of these systems, for example CORBA, can be implemented in other languages. The designed experiment is among other things going to answer how long time is needed to implement a specific application, how fast a specific client-server system is, and how long time is spent for learning about a specific system.</p>

The aim of this work is to design an experiment, where client-server systems can be compared. They belong to the group of systems that are called component systems. Client-server systems are difficult to compare, because they are complex. The client-server systems are documented in different ways. Notations in the implementation of them are in different ways. There is a large difference in the communication between the client and server. The architectures between the client-server systems differ also, but they are not totally different. Therefore it is possible to construct an experiment for comparing them. Client-server systems that will be compared are Java RMI, Web Services, CORBA, and Enterprise JavaBeans. We are going to use Java as the programming language. Some of these systems, for example CORBA, can be implemented in other languages. The designed experiment is among other things going to answer how long time is needed to implement a specific application, how fast a specific client-server system is, and how long time is spent for learning about a specific system.

風險管理已成為銀行重視的重要議題，而要做好風險管理，快速且正確地提供必要的資訊給管理者是必備的能力，而銀行的資訊系統架構龐大，重要的資訊散佈不同的資訊系統，但不同的資訊系統常無法整合以提供資訊，造成風險管理相關資訊取得不易或不符合時效性的要求。 本研究是從源頭的資料分析開始，領域則是限定於衍生性匯率金融產品，目的在於找出該領域商業概念中重要的性質，套用模型驅動架構的概念，利用IBM之資訊框架對商業概念進行分析整理，建立在該領域的本體論，並發展資料存取層之儲存綱要(schema)，本體論具備擴充性高及不易變動的穩定特色，可解決因使用者需求導致資訊系統架構變動頻繁的問題，最後再依循EJB架構開發商業概念之Entity Bean元件，完成從知識蒐集、知識分析到知識實現的整個知識管理過程。 / Risk management has become a critical issue that banks value much. To bring out great risk management, being able to provide managers with rapid, correct and essential information is the basic capability of all banks. However, due to the immense information architecture in which important information spreads over those information systems, the integration among different information systems become so unrealizable that managers usually receive insufficient or ineffective information. This study starts from the data analysis of the origins, and the study field is restricted to derivatives. The purpose of this study is to: 1. find out the key elements of the business concept within such research field; 2. by using IBM Information Framework, apply the concept of Model Driven Architecture to analyze the business concept. 3. via the Entity Bean of the EJB architecture, fulfill and accomplish the whole knowledge management processes, from knowledge collecting, knowledge analyzing to knowledge realization.

<p>Device Management Applications are applications which are used to manage software on devices such as mobile phones. OMSI Forum provides such an application which is used to update the software on a phone. Software updates can be done at device management stations placed in stores or other service locations. Whenever a phone's software is updated, information about the update process is stored in a log. These logs can then be analyzed to generate statistics about updates such as the number of successful or failed updates or which faults that are common.</p><p>This master thesis project solves the problem of manually collecting and compiling logs from several stores by making this process automatic. Rather than collecting logs manually, each device management station sends its logs to a centralized server which stores all collected logs in a database. This makes it possible to generate charts and statistics in a simple manner from a web application. This solution makes the analysis more e ective, allowing users to concentrate on analyzing data by removing the work task of collecting logs.</p>

Device Management Applications are applications which are used to manage software on devices such as mobile phones. OMSI Forum provides such an application which is used to update the software on a phone. Software updates can be done at device management stations placed in stores or other service locations. Whenever a phone's software is updated, information about the update process is stored in a log. These logs can then be analyzed to generate statistics about updates such as the number of successful or failed updates or which faults that are common. This master thesis project solves the problem of manually collecting and compiling logs from several stores by making this process automatic. Rather than collecting logs manually, each device management station sends its logs to a centralized server which stores all collected logs in a database. This makes it possible to generate charts and statistics in a simple manner from a web application. This solution makes the analysis more e ective, allowing users to concentrate on analyzing data by removing the work task of collecting logs.