Distribution

The benefits expected from distribution are commonly listed as:

Accessibility and availability of resources

The unity of the system must be preserved. Let's insist on the duality: availability here and now.

Scalability

Every system imposes limits (resources, performance) to scalability. Distribution is a way to overcome these.

Fault-tolerance

Distribution allows redundancies, which may be used to support resilience to local failures, or cross-checking of automated decisions.

Flexibility

Distribution offers potentialities for local maintenance, backup or updates, even at run-time. This is valuable e.g. for systems that cannot be disconnected from resources that they manage, for fear of loosing track of them.

Efficiency

Concurrency may, in many cases, improve performance.

These may be desirable for themselves, directly, but also indirectly, in order to reach some other goals, such as:

• Extensibility, which implies in practice scalability
• Robustness, which can be served by fault-tolerance
• Reusability, requiring accessibility and scalability
• Modularity...

The list given above is not unique. Its components are not orthogonal either: a correlation between the size of a system and loss in efficiency will set an upper limit to scalability, for instance.

Nevertheless, this list provides, better than a definition would, with a specification for distribution.

Table of contents