Single coherent system justification

Justification

A distributed system is a collection of independent computers, connected by a network and running distribution middleware, that appears to its users as a single coherent system. Its components communicate and coordinate their actions only by passing messages (Tanenbaum; Coulouris).

This statement means the many independent computers, the network and the replication behind it are hidden so that users and programs see a single, consistent computer.

Features that support this view

1. Transparency (access, location, migration, replication, concurrency, failure): the user uses one name space and one interface.
2. Middleware: a common layer between applications and the heterogeneous OS and network gives a uniform programming model.
3. Resource sharing: all authorised users share files, printers and services as in one system.
4. Consistency and concurrency control: concurrent users see a consistent state.
5. Fault tolerance: replicas and recovery hide crashes, so the service seems always available.
6. Scalability and openness: machines can be added without a change in the user's view.

Examples

• World Wide Web: a user types a URL and sees a page without knowing the server, its location or caches/mirrors behind it.
• Google Drive / Gmail: data is spread and replicated over data centres, but looks like one account.
• NFS: remote directories appear as part of the local file tree.
• Online banking: an account can be used from any branch or ATM as if it were one bank computer.