No description available.

No description available.

No description available.

Current Internet based auction services rely, in general, on a centralised auction server; applications with large and geographically dispersed bidder client bases are thus supported in a centralised manner. Such an approach is fundamentally restrictive as too many users can overload the server, making the whole auction process unresponsive. Further, such an architecture can be vulnerable to server's failures, if not equipped with sufficient redundancy. In addition, bidders who are closer to the server are likely to have relatively faster access to the server than remote bidders, thereby gaining an unfair advantage. To overcome these shortcomings, this thesis investigates ways of enabling widely distributed, arbitrarily large number of auction servers to cooperate in conducting an auction. Allowing a bidder to register with anyone of the auction servers and place bids there, coupled with periodic exchange of auction information between servers forms the basis of the solution investigated to achieve scalability, responsiveness and fairness. Scalability and responsiveness are achieved since the total load is shared amongst many bidder servers; fairness is achieved since bidders are able to register with their local servers. The thesis presents the design and implementation of an hierarchically structured distributed Internet auction system. Protocols for inter-server cooperation are presented. Each server may be replicated locally to mask node failures. Performance evaluations of centralised and distributed configurations are performed to show the advantages of the distributed configuration over the centralised one.

The adoption of the "Web Services" model for building a Grid framework created a considerable shift from the original concept of Grid which was based on "distributed job scheduling". The requirement for the access and integration of heterogeneous data resources over the Grid, and the advances in service-oriented data access standards led to the development of a service-oriented distributed query processor, which forms the basis of this thesis. The adoption of service-orientation raised the need for a framework which would allow demanddriven deployment of Web Services on available resources. Research into such concepts led to the development of DynaSOAr, a framework which proposed an alternative approach to distributed job scheduling by focussing entirely on the concept of services, rather than the more traditional jobs. DynaSOAr allows services to be deployed on demand to meet changing performance requirements and exploits the advances made in virtualization technologies to support the deployment of services and databases. The thesis describes a system designed to exploit dynamic deployment features within the context of distributed query processing on the Grid, and argues that such features benefit query evaluation by creating a loose coupling between the services and the available resources. The extended distributed query processing system is able to collocate various entities, such as the evaluation and analysis services with the data, to reduce data traffic over the network, and is also able to reconfigure itself to improve performance by dynamically deploying database snapshots. The thesis evaluates the dynamic distributed query processing framework through several experiments which explore its behaviour in a variety of scenarios.

No description available.

The aim of this work was to conduct experiments to discover the characteristics of the network traffic generated by running Grid applications. In the experiments, the Grid application used was to discover resources and services registered in non-central resource. The Vega Grid was introduced as the experimental Grid platform and Resource Discovery was run in this platform. Since the application of Resource Discovery could generate continuous network traffic, it was useful to measure and analyse this network traffic and find out its characteristics. Several experiments were conducted for the same purpose using three, four, five, and six PCs in different experiments. Moreover, the same experiment using a WAN was conducted with seven PCs (four PCs inside the campus and three PCs outside). Specifically, ETHEREAL was introduced to collect data packets involved with the experiments.

Grid computing is highly dynamic in nature where resources are subject to change due to performance degradation and node failure. The resources include processing elements, storage, network, and so on; they come from the interconnection of parallel machines, clusters, or any workstation. One of the main properties of these resources is to have changing characteristics even during the execution of an application. Thus, resource usage by applications cannot be static during run-time; neither can change in resources be considered as faults. Therefore, Grrid application designers must keep in mind that resources and resource management are highly dynamic within Grid architectures. Grdi resource brokering is introduced to simplify resource discovery, selection, and job submission for Grid application. However, it is the responsibility of a Grid resource broker to distribute jobs among heterogeneous resources and optimise the resource usage. As a result, a Grid resource broker should have the capablility to adapt to these changes and take appropriate actions to improve performance of various computing applications. To adapt to the Grid resource changes, an adaptive service is introduced in this research. The adaptive service consists of a monitoring tool, decision manager, and migration engine to ensure the job finishes at the time specified. The adaptive service supports job migration during run-time to ensure timely job completion. Our work in this research shows a Grid test-bed and White Rose Grid implementation of an adaptive service that supports job migration during run-time to ensure timely job completion. Performance prediction is used to estimate expected job completion time and determine whether any onserved performance degradation is likely to result in failure to meet a user specified deadline. A key feature of our approach is that the user is not required to install additional software or make complex alterations to their code requiring specialist Grid computing knowledge. This is achieved using a reflective technique to bind the adaptive service components to the user's code. Also, this research proves the adaptive service overhead is very minimal. The adaptive service is a viable contender for future Grid resource brokering implementation.

Grid based systems have increased the opportunity for users to deploy and execute their applications using Grid resources. These resources have varying reliability and performability, particularly when demand is high. If a Grid application is executed at such times, performance may suffer and results may be delayed. In order to overcome this problem, application management is needed to support Quality of Service (QoS) requirements. The Distributed Aircraft Maintenance Environment (DAME) is an example of a Grid based system in which users wish to attach application QoS requirements. In light of this, an adaptive SLA (Service Level Agreement) management system is presented which has the ability to interpret application requirements and deliver management using application adaptation. An SLA specification is presented which improves contract non-repudiation by way of elements which allow requirements, guarantees to be specified and provenance to be recorded. To predict the execution time of an application, a technique using historical observations is proposed. An approach which is highly appropriate for Grid based systems which perform countless runs of the same application. This prediction is used in combination with application monitoring to determine the progress made by the application during run-time. Progress is determined by comparing an estimate of the applications remaining execution time and an execution schedule. If application progress is insufficient, a rule-based control algorithm monitors progress and infers control actions which adapt the behaviour of the application. Experimental analysis is conducted on a local Grid test-bed and a large scale Grid infrastructure, the White Rose Grid. This shows the solution supports application executions with attached time or performance constraints; where use of the system prevents application failure or delay. Migration is useful in reducing the execution time of applications when performance degradation occurs. Mechanisms for automated monitoring and provenance capture are presented, both of which support the operation of the SLA management system. Adaptive SLA management benefits the users of Grid based systems such as DAME, by providing Grid application management. This is in contrast to current best-effort provision which offers no such guarantee. The ability to provide these guarantees and an SLA specification makes commercial exploitation of these Grid based systems more realistic

Scientific visualization today usually involves scientists from multiple disciplines trying to tackle complex simulation problems or dealing with large datasets. These scientists are likely geographically distributed, so there is a cost in both finance and time, when they are required to collaborate or meet face-to-face. The complex problems and larger datasets require high performance computing facilities which are not always available from these scientists’ desktops. The objective of this research is to address these two issues. We aim to create a novel visualization framework which supports collaborative working amongst different users and utilizes high performance computational resources in a Grid environment. The system is named as NoCoV, short for Notification-service-based Collaborative Visualization system. The system involves a number of visualization notification services which can be jointly provided by different developers and deployed on different Grid networks. A visualization notification service can subscribe to other services and receive notification messages delivered from them. By using this subscription/notification communication pattern, users can link suitable visualization services together through a centralized controller service to create customized visualization pipelines according to their different requirements. Collaborative working is supported in both creating and steering of visualization pipelines. Visualization information is recorded as XML-based descriptions and published as notification topics on a controller service. By subscribing to this controller service, collaboration can be achieved by sharing the visualization information amongst different participants. The NoCoV system accommodates users with different background knowledge and supports collaboration between them. Visualization services are jointly provided by different visualization service developers. Visualization experts can compose pipelines with the services created by service developers using their visualization expertise. Application scientists can use pipelines created by visualization experts and can set steering parameters according to their specialized knowledge. A NoCoV prototype is implemented as proof of concept. The prototype has four visualization services and one controller service which are implemented as WSRF notification services deployed on GT4 containers. The prototype also provides a pipeline editor client GUI for visualization experts to create visualization pipelines and a parameter control client for application scientists to steer these pipelines. The performance of the NoCoV prototype is evaluated by a test involving a pipeline distributed across different Grid networks. The limitations of the NoCoV system and the future work are also discussed at the end of this thesis. The overall vision of the NoCoV system is to produce the next generation of visualization system, which has a worldwide repository of visualization services deployed on different Grid networks and supports collaboration amongst users on a global scale.