Distributed System

A distributed system is a collection of autonomous computing elements(also known as node) that appears to its user as a single coherent system.

Problems:

• Each node has its own time, no global clock
• Group membership management and authorization

Overlay network is the organization of the distributed system. Nodes only communicates with its neighbours. There are 2 types of overlay network: structured and unstructured.

Example: peer-to-peer system

Goals:

• Distribution transparency
• Partial failures transparency

Middleware is the commonly used components and functions for every nodes.

Transparencies

1. Access: Hide data representations of objects.
2. Location: Hide physical/logical location.
3. Relocation: Move the object when it is in use.
4. Migration: Move the object.
5. Replication: Hide replicas of the object.
6. Concurrency: Allow the object used by several indipendent users.
7. Failure: Hide failure and recovery of an object.

Openess

1. Comform to a well-defined interface
2. Support portability of applications
3. Make extensibility
4. Easily interoperate

Scalbility

1. Size: including storage, computation, and bandwith resource
2. Geographical
3. Administrative