distributed systems architecture

Distributing a system over several processing nodes means, in general, that
(benefits):

• things happen concurrently, so you can get more done
• by replicating functionality, you can provide a more robust service
• users and other actors in different locations can readily be served
• not all the functional nodes need be up at once
• diverse systems may be connected

and (obstacles to design):

• things happen concurrently, admitting interference, deadlocks, etc
• different nodes may get out of sync
• communication between the nodes may be unreliable, slow, and erratic
• not all the nodes may be up at once
• the execution environments running the distributed parts  may differ (different languages, operating systems, performance, resources available, ...)

Distributed systems can be open or closed or in-between. Very open systems like the Internet have no-one in control: broad standards of communication are set, which are often extended for particular purposes by different groups of users. In a closed system, the designers have control over the whole lot, though they usually want it to be readily extensible: the different parts of an aircraft control system are an example.
Besides the Internet, various large systems (some of which are built atop the Internet) fall into the open category: for example, a system (described by XXXX at OOPSLA98) for retrieving medical records from the patient's home medico and various places he may have been treated, to a hospital anywhere in the world. 

complete solutions for e-business
(consultancy, courses, workshops, mentoring, seminars, development)