Tim Kindberg

Sharing objects over the Internet: the Mushroom approach

This paper reports on the Mushroom project. The project is developing a software framework for collaborative working and user interaction on the Internet. The paper illustrates some of the required features in the context of an application scenario and outlines those aspects of the Mushroom system architecture that support the sharing of information. The projects motivation is to address the problems of coordinated user interaction, distribution and privacy. Mrooms are interactive environments for groups of collaborating users. They provide a shared space in which users are aware of one another while working on shared objects. Their boundaries provide a triggering mechanism for consistency and privacy checks. In contrast to other systems with room-based metaphors, Mushroom focuses on a scalable and flexible system architecture using replicated state, group communication and event-based updates.

A QoS support framework for dynamically reconfigurable multimedia applications

The use of multimedia in distributed systems has begun to include such complex and mission-critical domains as digital television production, ‘video-ondemand’ services, medical and security systems. These applications impose more stringent requirements on the support mechanisms provided by underlying networks and operating systems than most currently deployed continuous media applications. This paper describes the Djinn multimedia programming framework, which is designed to support the construction and dynamic reconfiguration of distributed multimedia applications. We motivate the benefits of a runtime model of the quality of service and other characteristics of multimedia applications, and demonstrate a generic algorithm for scheduling dynamic reconfigurations that maintains QoS guarantees. QoS characteristics are modelled as piecewise-linear or quadratic relations, which are solved using standard constraint programming techniques. During reconfigurations, updates to active components are scheduled so as to maintain temporal constraints on the media streams. We illustrate our approach using experimental results from a real-world application domain.

Dynamically reconfiguring multimedia components: a model-based approach

Distributed multimedia systems are potentially subject to frequent and ongoing evolution of application structures. In such systems it is often unacceptable for reconfigurations to fail or to only partially succeed. This paper describes the reconfiguration architecture of the DJINN multimedia programming framework. We introduce the concept of multimedia transactions for structuring changes into atomic units that preserve application consistency and extend these with the smoothness condition to maintain temporal as well as data integrity across reconfigurations. We present a technique for scheduling configuration changes that trades off the perceived level of smoothness against the available resources and the desired timeliness of the reconfiguration.

Adaptive parallelism under Equus

The authors describe adaptively parallel computations under Equus (T. Kindberg, 1990; A. V. Sahiner, 1991). These computations execute on a processor pool, and expand and contract as the number of processor nodes allocated to them varies over their run-time. They are based upon a hierarchical master-worker structure. The number of worker processes changes with the number of allocated nodes, and so does the number of processes that act as servers to them (such as the masters). The authors use an image-processing example to describe how workers and servers are added and withdrawn at run-time. The affected processes are synchronised, communication linkages are changed, and in some cases, state is transferred between them. Reconfigurations are transparent to worker (and other client) processes, but not all can be made transparent to servers.<<ETX>>

A sequencing service for group communication

This paper is about the design of a sequencing service for totallyand causally-ordered group communication in a distributed system. A group communication service provides facilities for managing groups of processes and for multicasting messages to all the members of a group. The primary role of the sequencing service is to impose a total order on multicast messages, but we shall show how to build a combined causal order from it, and also a new hybrid order which we call cautal ordering. This design addresses the issue of sequencing many groups, which may overlap, and of using several sequencer processes to do so. In the remainder of this section we briefly explain the requirement for group communication involving overlapping groups, in the context of multi-user applications; we describe existing sequencer-based protocols for totally-ordered multicast to a single group, and we summarise our main contributions in relation to them. Section 2 describes extensions to sequencer-based protocols to handle multiple groups, and to split the sequencing load between several sequencers whenever possible. Section 3 shows how to obtain causal and cautal ordering from total ordering. Section 4 discusses our protocols, and summarises and concludes the paper.

Security for Network Places

A network place is a persistent environment in which a collection of networked users can share information objects and communicate with one another. For example, a group of collaborating clinicians can use a network place to manage a patient’s records. Much of the information that users store in places and their communications within them are confidential: users require that they be kept secret, and that the integrity of their information is maintained.