Multimedia applications of the future will involve the distributed processing of multimedia information in systems ranging from large servers, through workstations all the way down to ``set top boxes''. The implementation of such applications will be greatly eased by the use of a common architecture providing application level quality of service, and, where possible, a common software platform even where the underlying hardware has diverse capabilities.
Pegasus II takes the results of the Pegasus project, performed in the BRA programme, and the DCAN project as a starting point on which to build a complete distributed multimedia platform, including toolkits, user interfaces, filing systems and the emulation of familiar software environments.
The work is based on the philosophy of the Pegasus project:
Pegasus II will employ the operating system kernel (Nemesis) and storage system built in the course of the original Pegasus project as a basis for future work, and will extend it to cover the outstanding issues which we have learnt are essential for industry acceptance of the Pegasus architecture: support of current applications and programming environments, and effective user control over the resource allocation policy via an obvious user interface.
The results of Pegasus will be used as the underlying operating system. Furthermore, an existing UK funded collaboration (project DCAN) between and Cambridge has already ported the -RT system to Nemesis to enable it to be used as a platform for distributed control of telecommunications networks.
Pegasus II will have as its first concern the higher level functionality: emulation of standard environments to enable porting of application software; toolkits for scheduling, synchronisation and manipulation of continuous media; and user interfaces which allow user control of resource management. However, more fundamental work in the development of the kernel and of networking technologies will also have a place in the work.
In short, the project will migrate from a successful research prototype to a system available to industry for evaluation and experimentation which provides many of the familiar features required to make its adoption possible.