This paper presents the basic ideas and the technical background of the Esprit project HiPEC. The HiPEC Project aims at integrating HPCN technologies into an electronic commerce scenario to ease the interactive configuration of products and their photorealistic visualization by a high performance parallel rendering system.

In this paper a new mechanism for mapping data items onto the storage devices of a parallel web server is presented. The method is based on careful observation of the effects that limit the performance of parallel web servers, and by studying the access patterns for these servers.On the basis of these observations, a graph theoretic concept is developed, and partitioning algorithms are used to allocate the data items. The resulting strategy is investigated and compared to other methods using experiments based on typical access patterns from web servers that are in daily use.

This paper presents the SICMA Multimedia Information Server that was demonstrated during the first public trial of the ACTS project SICMA, which took place at the Natural History Museum of London and in the Gallo Romeins Museum of Tongeren (Belgium), from June to September of 1997. The demonstrator application, namely the “Virtual Museum”, is also presented. The SICMA server complies to the DAVIC standard and is built on top of the KYDONIA multimedia information server. Core DAVIC Server Service elements (Service Gateway, Stream, Content, File) have been developed and the DAVIC User Plane information flows (S1 and S2) are supported. In this paper we describe the overall SICMA server architecture and explain the main factors that lead to server design. We also identify the critical points which affected the overall system performance and determined the degree of compliancy with the accepted standards, such as DAVIC.

In this article, we present an implementation of a distributed branch and bound algorithm solving the Vertex Cover problem on a network of up to 63 Transputers. We implemented two different strategies: The first parallelization of our branch and bound algorithm is fully distributed. Every processor performs the same algorithm but on a different part of the solution tree. In this case it is necessary to distribute subproblems among the processors to achieve a well balanced workload. Our second strategy is based on a tree structured network, where all subproblems are stored at the root processor and the other processes work as slaves of this master process. To show the performance of our strategies, we solved the Vertex Cover problem for graphs of up to 150 nodes and an average degree of 30. We were able to achieve a speedup of 57.35 for the first strategy on a 60 processor network and 62.11 for the second strategy on 63 processors, compared to a very efficient sequential algorithm.

This paper presents a new algorithm for the detection and resolution of communication deadlocks in distributed systems. The algorithm is based on some well known concepts for distributed deadlock detection and adds some new features to reduce message- and space complexity. It was implemented on a transputer network and shown to be more efficient than previously published algorithms.

In this paper, we describe an architecture for a scalable distributed server management system which is capable of managing media assets stored on large-scale media server networks. In addition to the functionality for a decentralized management of the whole server network, the server management system allows to make bandwidth reservations even in advance for streaming in real-time and transmitting media assets. This increases the QoS for end-users since bandwidth reservations for streaming applications can be planned and established a long time before the actual streaming process takes place. Moreover, this approach allows a tight estimation of the duration required for copying media assets among different servers of the network. In order to minimize the required amount of network bandwidth for copying media assets and to reduce the total number of copying processes, the management software provides the functionality to place media assets on the servers of the network in a way that assets of particular interest for a certain client are located on the client’s local server. In this paper, the architecture of the server management system itself is presented together with the approach used to provide QoS guarantees for end-users. The overall system is using the differentiated service model. Besides the technical concept we also present results of the integration of the system into the corporate network of a german company.

We present an overview of a large combined querying and retrieval system that performs content-based on-line searches in a large database of multimedia documents (currently text, tables and colour images). Queries are submitted as sentences in natural language and are transformed into the language of the target database. The documents are analyzed semantically for their information content; in a data fusion step the individual pieces of information extracted from these documents are aggregated into cognitively adequate result documents. There is no pre-indexing necessary when new documents are stored into the system. This retains a high degree of flexibility with respect to the questions that may be asked. It implies, however, that both huge amounts of data must be evaluated rapidly and that intelligent caching strategies must be employed. It is therefore mandatory that the system be equipped with dedicated high-speed hardware processors. The complete system is currently available as a prototype; the paper outlines its architecture and gives examples of some real sample queries in the knowledge domain of weather data documents.

In this paper we propose a hierarchical server network architecture for the support of a large-scale TV-Anytime system. We present and formulate various policies used for the effective mapping of videos in the runtime of the TV-Anytime system. A predictive assignment policy is presented to determine periodically the mapping of videos to the server network. Furthermore, we present an adaptive reallocation policy to handle short-term variations. Our simulation results demonstrate that these policies yield an efficient system and make the system more adaptive to changes in video popularity and system state. Our simulation results also show that batching and multicasting significantly improves the system throughput during high load periods and makes the system more scalable.

Advances in storage and high-speed broadband communication,network technologies are making it feasible to design and implement distributed scalable Video-on-Demand structure networks. In this paper; a hierarchical structure for a server network is proposed to support a metropolitan TV-Anytime service. A key issue, how to map the media assets onto such a hierarchical sen,er network, is addressed in this paper: The definition and heuristic methods for the solution of this new combinatorial optimization problem are presented. The mapping problem studied here addresses the management of media assets taking the storage capacity: communication bandwidth limitation and access pattern into account. The methods presented in this payer are based on parallel simulated annealing and are verified by the use of a set of benchmark instances. The results show that the methods provide a near to optimal performance in terms of the Quality of Service (QoS).

This paper presents algorithms and principles for the implementation of a novel broadband multimedia service called "TV-Anytime (TVA)". The basic idea of this service is to store broadcast media information distributed by TV broadcasters on media server systems that are installed close to the later user and allow this user to access the stored broadband media information at any time. Using a given user profile the recording of media assets can be personalized allowing the implementation of "Personal TV" in this way.The paper studies the implementation of TV-Anytime services on scalable server networks that incorporate local area networks, wide area networks as well as last-mile networks like ADSL and CATV. The mapping of media assets on a network of distributed media servers is studied here. We present methods that take up concepts from graph for the management of media assets and verify these methods using sample applications. The-results show that the methods provide near to optimal performance in terms of the "Quality of Service" for the TVA scenario.

This paper presents principles and algorithms for the management of a network of media servers. Such a server network allows the online delivery of broadband multimedia data, e.g. audio and video streams, to a large number of widely distributed clients. Thus, the implementation of large scale distributed broadband media information services is possible if such networks can be handled efficiently. In the paper two new combinatorial optimization problems are defined that have to be solved for an efficient management of the distributed media servers connected by a high-bandwidth communication network. The algorithms consider the case of a static mapping of media assets onto the server network as well as the dynamic migration of media assets on such a network. The first scenario is used in the case that all media assets are loaded for the first time onto the server network. The second scenario is applied dynamically during run-time of an information service taking the varying user access pattern and changing media content of the server network into account. In this paper the algorithms are presented and their efficiency is demonstrated using some benchmark instances. The algorithms are part of a larger distributed system that allows the management of a network of distributed media servers. The principles of this system are presented as well as some applications that have been developed and are supported by this distributed server management system (DSMS)

This paper presents a methodology for the efficient implementation of large scale distributed broadband multimedia information systems. The principle presented in this paper can be applied to a large number of different application scenarios. The technical basis is a novel network of media servers providing audio and video streams of high quality to a large number of widely distributed clients. The server network is managed by a distributed server management system presented in this paper. To provide media streams to widely distributed users intelligent caching and mirroring mechanisms are used that are based on user profiling and content classification.

The provision of interactive multimedia services, such as video-on-demand, teleshoping and distance learning, to a large number of users, still remains a challenging issue in the multimedia area. Despite of recent technological advances in all levels of the distributed multimedia infrastructure (storage, network, compression, standardization etc..), there is a strong need for feedback from public trials. Trial descriptions and evaluations, by revealing potential system limitations and measuring end user reactions, will provide valuable input towards the large-scale deployment of such services. In this paper, we present the teleteaching trial that was held at the end of 1998, at Limburg University (Belgium), in the context of ACTS SICMA project. We describe in detail the overall architecture and present results/implementation experiences for the parts of the system, putting the main emphasis on the server. We present technical integration issues between DAVIC and Internet server protocols (RTSP, DSM-CC etc..), and discuss the overall trial results.

The paper presents algorithms and principles for the management of large scale broadband multimedia services. For the implementation of these services, we use a network of distributed media servers, storing broadband media information e.g. audio and video that are streamed from the server to the connected clients. The problem studied in the paper is the managing of media content on a server network. We present the Distributed Server Management System (DSMS) that performs an efficient content and service management on the distributed servers. The DSMS allows one to collect access patterns from the users of the multimedia services and uses this knowledge for the solution of some combinatorial optimization algorithms that have to be solved for an efficient assignment of media assets. We present these optimization algorithms and evaluate their performance using some benchmark instances.

The paper presents a mechanism for mapping data items onto the storage devices of a parallel Web server. The method is based on careful observation of the effects that limit the performance of parallel Web servers and by studying the access patterns for these servers. On the basis of these observations, a graph theoretic concept is developed and partitioning algorithms are used to allocate the data items. The resulting strategy is investigated and compared to other methods using experiments and typical access patterns from Web servers that are in daily use

The authors consider the simulation of large cube-connected cycles (CCC) large butterfly networks (BFN) on smaller ones, a problem that arises when algorithms designed for an architecture of an ideal size are to be executed on an existing architecture of a fixed size. The authors show that large CCCs and BFNs can be embedded into smaller networks of the same type with (a) dilation 2 and optimum load, (b) dilation 1 and optimum load in most cases, (c) dilation 1 and nearly optimum load in all cases. The results show that large CCCs and BFNs can be simulated very efficiently on smaller ones. Additionally, they implemented their algorithm for compressing CCCs and ran several experiments on a Transputer network, which showed that the technique also behaves very well from a practical point of view.<>

The paper describes the design of a scaleable interactive continuous media server that is realized using a distributed memory parallel computer architectures. We present a scheme for data layout, admission control and communication scheduling that provides inherent Quality of Service support for the delivery of continuous media streams from the storage devices to the external communication interfaces within the internal communication network of the media server. The methods presented here are used for the development of a parallel continuous media server that is integrated with different applications and supports the RTP/RTCP protocol for data delivery as well as the RTSP protocol to control the delivery of media streams.

Heterogeneous parallel computing systems built up by closely connected networks of workstations or personal computers are a cheap alternative to dedicated parallel supercomputer systems. As these systems are widely available in academic and industrial environments it is becoming increasingly popular to use these resources to solve time consuming calculations.In this paper, the question of load balancing on heterogeneous parallel computing systems is addressed. The paper presents a general methodology to transform existing load balancing methods that are used for load balancing on homogeneous parallel computing systems into methods, which tate into account the special characteristics of heterogeneous systems.The here presented strategy is applied to a load balancing algorithm, which has been used for distributed branch & bound applications on homogeneous parallel computing systems. Experiments conducted on a 16-node ATM cluster show that the speedup of the parallel branch & bound application can be increased considerably by the use of a transformed algorithm.

In this paper we discuss the design of parallel interactive continuous media servers suitable for the implementation of scalable server-based media delivery services like Video-on-Demand or Teleshopping. The main design problems for the development of such servers is to ensure the just-in-time delivery of media elements in order to maximize the Quality of Service and to minimize the buffer size at the user site. Just-in-time delivery means that the media elements should be sent as late as possible to the users but early enough to ensure a continuous replay of the media. This is important because clients have to provide buffer space for data arriving to early. The Quality of Service measures the number of data elements arrived in time at the user side. Thus, the real-time properties of the internal communication network as well as the congestion arising at the disks are of highest importance. We present models for parallel media servers and a very simple scheduler that is fully distributed and can therefore easily be implemented on a scalable parallel continuous media server. For each requested data element the scheduler sends a request to the storage subsystems at a point of time only depending on the deadline of that request, i.e. the time the data has to be delivered to the user, and the length of the path the data has to be routed through the internal network of the parallel server. In order to minimize the buffer space at the user site, and to maximize the Quality of Service, we develop timing strategies for the scheduler using simulation results as well as analytical observations.

This article focuses on principles for the design of efficient parallel algorithms for distributed memory computing systems. We describe the general trend in the development of architectural properties and evaluate the state-of-the-art in a number of basic primitives like graph embedding, partitioning, dynamic load distribution, and communication which are used, to some extent, within all parallel applications. We discuss possible directions for future work on the design of universal baste primitives, able to perform efficiently on a broad range of parallel systems and applications, and we also give certain examples of specific applications which demand specialized basic primitives in order to obtain efficient parallel implementations. Finally, we show that programming frames can offer a convenient way to encapsulate algorithmic know-how on applications and basic primitives and to offer this knowledge to non-specialist users in a very effective way.