Distributed Computing Algorithm
The area of simultaneous and distributed computing studies similar questions in the case of either multiple PCs, or a PC that executes a web of interacting processes : which computational issues can be cleared up in such a network and how efficiently? it isn't at all clear what's meant by unravelling a problem in the case of a simultaneous or distributed system : for instance, what's the task of the algorithm designer, and what's the simultaneous and / or distributed equivalent of a sequential general-purpose computer? The conversation below focusses on the case of multiple computers, though plenty of the problems are the same for simultaneous processes running on a single PC.
Three perspectives are typically used:
Parallel algorithms in shared-memory model
* All PCs have access to a shared memory. The algorithm designer selects the program executed by each PC.
* One unproven model is the parallel random access machines ( PRAM ) are used. the classical PRAM model presumes synchronous access to the shared memory.
* A model that's nearer to the behaviour of real-world multiprocessor machines and accounts for the utilization of machine instructions such as Compare-and-swap ( CAS ) is that of asynchronous shared memory. There's a wide body of work on this model, an outline of which can be discovered in the literature.
Parallel algorithms in message-passing model
* The algorithm designer selects the structure of the network, as well as the program executed by each PC.
* Models such as sorting networks and Boolean circuits are used. A Boolean circuit can be thought of as a PC network : each gate is a PC that runs an intensely easy PC program. Likewise , a sorting network can be thought of as a PC network : each comparator is a PC.
Distributed algorithms in message-passing model
* The algorithm designer only selects the PC program.
All PCs run the same program. The system must work properly without regard for the structure of the network.
* A usually used model is a graph with one finite-state machine per node. In the case of distributed algorithms, computational issues are typically related to graphs. Regularly the graph that describes the anatomy of the PC network is the issue example. This is illustrated in the following example.
An example
Let consider finding a coloring of a given graph G. Different fields might take the following approaches :
Distributed algorithms
* The graph G is the anatomy of the PC network. There is one PC for each node of G and one communication link for each edge of G. At first , each PC only knows about its swift neighbors in the graph G ; the PCs must exchange messages with one another to find out more on the structure of G. Each PC must produce its own color as output.
* the target is on coordinating the operation of a capricious distributed system.
While the sector of parallel algorithms has a different focus than the field of distributed algorithms, there's a lot of communication between the 2 fields. As an example, the ColeVishkin algorithm for graph coloring was initially presented as a parallel algorithm, but the same system can also be used directly as a distributed algorithm. Furthermore , a parallel algorithm can be implemented either in a parallel system ( using shared memory ) or in a distributed system ( using message passing ) .[28] The standard line between parallel and distributed algorithms ( select an appropriate network vs. Run in any particular network ) doesn't lie in the same place as the borderline between parallel and distributed systems ( shared memory vs. Message passing ).
