|
Distributed Computer Systems have become competitive in providing
large amounts of computational power at a very low cost. The system
consisting of personal computers, mini or main frames, or high
performance multiprocessors, integrated into a high speed computer
network is capable of providing any organization the power of a
super machine with only a small initial cost. Such a system has the
further benefit of providing industry with an easy and modular
upgrade path because increasing the power of the system simply
involves increasing the number of networked computers. Distributed
Multimedia Applications (DMA) are so rich in their diversity of
methodology and so inherently computational intensive that they
naturally require a very heterogeneous mix of distributed processors
interconnected by an efficient interconnection network. A DMA can be
represented by a precedence graph, where nodes represent components
(or modules) of the application, interconnected by arcs representing
data streams flowing between different components. In this paper we
study the problem of partitioning and mapping Distributed Multimedia
Application graphs onto a heterogeneous distributed computer system.
Such applications require a compromise between quality of service
and cost of the utilized resources of the distributed computer
system. As the problem is difficult to solve, in general, we present
an approximate scheme which optimally assigns task modules of the
application onto the processors of the distributed system.
|