Objects and processes are often confused in the literature. An
instance of an implementation of an object <#209#> may<#209#> be a process (in the
Unix sense of the word).
However, there is no real reason to prevent the implementor
programming many objects in a single process. From the other extreme,
at some level (view), a collection of processes may implement a single
object.
However, in Distributed Systems, we have true concurrency, and
separate address spaces between different machines. In this case, it
may be wise to implement objects as processes.
Reasons include:
-
Robustness
-
Performance
-
Availability
This notion is discussed at greater length in Chapters 2 and 3.