7/20/14

Group Communication in Distributed Systems, AI.

one-to-one message exchanges are useful, but often not enough.

to communicate something by one process to a group of processes, we use multicast messaging.

in the context of Distributed Applications, it has following uses:

* Fault Tolerance: by having & using 'reserve services',
* Finding Objects in Distributed Services: by sending a message to many services at once, with a query, and waiting for answers,
* Performance increase, sometimes: by data multiplication. sometimes data is closer (less hops in the Internet, for example), or we just have less systems that are overburdened, perhaps more,
* Multicast Updates: sometimes we can notify many systems at once of an event occurance,
* perhaps more.

above have uses in a Service Integration or Systems Integration (putting the pieces together so they work well together).

Systems Integration: Service can be provided by a Computer System (hardware with software),

Services Integration: multiple Services can be Integrated within a single Computer System, but not only within a single Computer System.

i think that Sitie Space with this post have uses in the Distributed Artificial Intelligence, for communication (sending messages) between 'Neurons' (machines such as Computer Systems), connected as in a Graph by the Internet. Each of Computer Systems can also consist of a 'Graph of Neurons and Connections', together providing a 'larger' Service of a more complex 'Neuron' - it's 'Recurrential'.

Source: [13], Wikipedia (links are above), my experience.

See also: Distributed System, What is cluster?, Petri Nets, Stitie Space and Distributed Machines, Simple Artificial Intelligence, Hardware, Operating System and Computer System (for everyone), Atomic Multicast, Reliable Multicast.

No comments:

Post a Comment