Currently, several mature and commercial component models (for e.g. EJB, .NET, COM+) exist on the market. These technologies were designed largely for applications with business-oriented non-functional requirements such as data persistence, confidentiality, and transactional support. They provide only limited support for the development of components and applications with non-functional properties (NFPs) like QoS (e.g. throughput, response time). The integration of QoS into component infrastructure requires among other things the support of components’ QoS contract specification, negotiation, adaptation, etc. This thesis focuses on contract negotiation. For applications in which the consideration of non-functional properties (NFPs) is essential (e.g. Video-on-Demand, eCommerce), a component-based solution demands the appropriate composition of the QoS contracts specified at the different ports of the collaborating components. The ports must be properly connected so that the QoS level required by one is matched by the QoS level provided by the other. Generally, QoS contracts of components depend on run-time resources (e.g. network bandwidth, CPU time) or quality attributes to be established dynamically and are usually specified in multiple QoS-Profiles. QoS contract negotiation enables the selection of appropriate concrete QoS contracts between collaborating components. In our approach, the component containers perform the contract negotiation at run-time. This thesis addresses the QoS contract negotiation problem by first modelling it as a constraint satisfaction optimization problem (CSOP). As a basis for this modelling, the provided and required QoS as well as resource demand are specified at the component level. The notion of utility is applied to select a good solution according to some negotiation goal (e.g. user’s satisfaction). We argue that performing QoS contract negotiation in multiple phases simplifies the negotiation process and makes it more efficient. Based on such classification, the thesis presents heuristic algorithms that comprise coarse-grained and fine-grained negotiations for collaborating components deployed in distributed nodes in the following scenarios: (i) single-client - single-server, (ii) multiple-clients, and (iii) multi-tier scenarios. To motivate the problem as well as to validate the proposed approach, we have examined three componentized distributed applications. These are: (i) video streaming, (ii) stock quote, and (iii) billing (to evaluate certain security properties). An experiment has been conducted to specify the QoS contracts of the collaborating components in one of the applications we studied. In a run-time system that implements our algorithm, we simulated different behaviors concerning: (i) user’s QoS requirements and preferences, (ii) resource availability conditions concerning the client, server, and network bandwidth, and (iii) the specified QoS-Profiles of the collaborating components. Under various conditions, the outcome of the negotiation confirms the claim we made with regard to obtaining a good solution.

Currently, several mature and commercial component models (for e.g. EJB, .NET, COM+) exist on the market. These technologies were designed largely for applications with business-oriented non-functional requirements such as data persistence, confidentiality, and transactional support. They provide only limited support for the development of components and applications with non-functional properties (NFPs) like QoS (e.g. throughput, response time). The integration of QoS into component infrastructure requires among other things the support of components’ QoS contract specification, negotiation, adaptation, etc. This thesis focuses on contract negotiation. For applications in which the consideration of non-functional properties (NFPs) is essential (e.g. Video-on-Demand, eCommerce), a component-based solution demands the appropriate composition of the QoS contracts specified at the different ports of the collaborating components. The ports must be properly connected so that the QoS level required by one is matched by the QoS level provided by the other. Generally, QoS contracts of components depend on run-time resources (e.g. network bandwidth, CPU time) or quality attributes to be established dynamically and are usually specified in multiple QoS-Profiles. QoS contract negotiation enables the selection of appropriate concrete QoS contracts between collaborating components. In our approach, the component containers perform the contract negotiation at run-time. This thesis addresses the QoS contract negotiation problem by first modelling it as a constraint satisfaction optimization problem (CSOP). As a basis for this modelling, the provided and required QoS as well as resource demand are specified at the component level. The notion of utility is applied to select a good solution according to some negotiation goal (e.g. user’s satisfaction). We argue that performing QoS contract negotiation in multiple phases simplifies the negotiation process and makes it more efficient. Based on such classification, the thesis presents heuristic algorithms that comprise coarse-grained and fine-grained negotiations for collaborating components deployed in distributed nodes in the following scenarios: (i) single-client - single-server, (ii) multiple-clients, and (iii) multi-tier scenarios. To motivate the problem as well as to validate the proposed approach, we have examined three componentized distributed applications. These are: (i) video streaming, (ii) stock quote, and (iii) billing (to evaluate certain security properties). An experiment has been conducted to specify the QoS contracts of the collaborating components in one of the applications we studied. In a run-time system that implements our algorithm, we simulated different behaviors concerning: (i) user’s QoS requirements and preferences, (ii) resource availability conditions concerning the client, server, and network bandwidth, and (iii) the specified QoS-Profiles of the collaborating components. Under various conditions, the outcome of the negotiation confirms the claim we made with regard to obtaining a good solution.

Die folgenden Fragen standen im Mittelpunkt der Dissertation: Wie handeln Grundschulkinder aus, wenn sie von ihren Peers ungerecht behandelt werden? Welche unmittelbaren Wirkungen hat ihr Vorgehen? In welchem Zusammenhang steht das Vorgehen in der Konfliktsituation mit der Stellung unter den Peers? Theoretische Grundlage waren die Entwicklungsmodelle zum Aushandeln von Yeates und Selman (1989) sowie Hawley (1999). Es nahmen 213 Dritt- und Fünftklässler an der Untersuchung teil. Eine Kombination qualitativer und quantitativer Methoden wurde verwendet. In einem individuellen Interview wurde erhoben, welche Taktiken (eine Handlungseinheit) sowie welche Strategien (Abfolge von Taktiken) die Kinder in einer hypothetischen Normbruchsituation einsetzen würden. Die Kinder wurden auch gefragt, welche unmittelbaren Wirkungen sie erwarten, wenn sie die vorgeschlagene Taktik einsetzen. Die Stellung der Kinder wurde sowohl bezüglich ihres Einflusses (Peerrating) als auch bezüglich ihrer Akzeptanz (Soziometrie) unter den Klassenkameraden erhoben. Die von den Kindern genannten Taktiken wurden vier übergeordneten Kategorien zugeordnet: Verhandeln, Erzwingen, Ausweichen und Aufgeben. Nach den Erwartungen der Kinder führen sowohl Verhandlungstaktiken als auch erzwingende Taktiken in um die Hälfte der Fälle zur Durchsetzung. Erzwingende Taktiken gehen jedoch häufig mit unfreundlichen Reaktionen einher. Einfluss und Akzeptanz der Kinder waren davon abhängig, welche Kombination und Sequenz von Taktiken (Strategie) sie wählten. Beispielsweise waren Kinder einflussreich und beliebt, die eine Reihe von Verhandlungstaktiken generierten oder die zunächst Verhandlungstaktiken vorschlugen und danach erzwingende Taktiken. Kinder, die sofort Zwang einsetzen würden, hatten wenig Einfluss und wurden abgelehnt. Außerdem fanden sich Geschlechts- und Altersunterschiede hinsichtlich des Vorgehens in der hypothetischen Normbruchsituation sowie der Zusammenhänge zwischen Taktiken und Stellung unter Peers. / The dissertation dealt with the following questions: How do children react, when a peer infringes upon their rights? What immediate outcomes are connected to their actions? What is the relation between the way children proceed in the conflict situation and their social standing among their peers? As a theoretical background served the modells about negotiations by Yeates and Selman (1989) as well as Hawley (1999). 213 third- and fifthgraders participated in the study. A combination of qualitative and quantitative methods was used. Children were individually interviewed to assess what tactics (unit of action) as well as what strategies (sequence of tactics) they would apply in a hypothetical conflict involving a rights infraction. Children were also asked what immediate outcomes they expected, using the proposed tactic. Children's social standing was assessed concerning their influence (peerrating) as well as their acceptance (sociometrie) among their classmates. The tactics propsed by the children were assigned to four superordinate categories: negotiate, coerce, get out of the way, give in. According to the expectations of the children both negotiation tactics and coercive tactics lead to assertion in about half of the cases. Coercive tactics however often evoque unfriendly responses. Children differed in their influence and acceptance depending on the combination and sequence of tactics (strategy) they chose. For example children, which generated a series of negotiation tactics and children, which proposed negotiation tactics in the beginning and coercion tactics later were both influental and liked. Children, which chose coercion allready in the beginning had little influence and were rejected. Also gender and age differences were found concerning the tactics chosen in the hypothetical conflict situation and the relation between tactics and social standing among peers.