Distributed Computing

Back in the day

Distributed used to mean computers that were spread out in a geographic area, but today can mean autonomous processes that run on the same computer, and interact with each other by message passing.

Today it means components located on connected/networked computers. Thes computers communicate and coordinate their actions by passing messages. These components interact to achieve a common goal.

Significant characteristics include concurrency of components, lack of a global clock, and independent failure of components.

Low level- CPUs are connected via some sort of network, be it printed onto a circuit board or made up of loosely coupled devices. Higher level- A communication system must interconnect these processes running on those CPUs.

A note on reliability. Distributed systems are more reliable. Many low end computers provide a lower chance of failure than one high end computer.

The subtle, and unclear distinctions

Concurrent computing, parallel computing, and distributed computing have a lot of overlap and no clear distinction. A system could be described as parallel and distributed. Distributed computing is a loosely coupled form of parallel computing ( parallel computing- tight coupling of distributed).

Distributed program is a distributed system that is run by a computer program. In distributed programming it is common for tasks to be partitioned out and solved by one, or more, computers.

Parallel computing means all processors have access to a shared memory to exchange information. Distributed computing means each processor has its own private memory, and information is exchanged by passing messages between processors.

The forms of distributed computing

Distributed programming falls into some of these categories:

Client-server

Clients contact the server for data then format and display into the users. Permanent changes are committed back to the server. In this model work is done by a server, and is requested by a client. The server does the work and the client initiates the contact.

Examples: email, network printing, the world wide web.

Tier applications

Three-tier- Most web applications. Architectures that move the client intelligence to a middle client. Simplifies application deployment.

N-tier- refer to web applications which further forward their requests to enterprise services. This lead to the success of application servers.

Multi-tier architecture - client server architecture presentation, application processing, and data management functions are physically separated. Three tier is the most common.

Examples: web development such as e-commerce website. A front end web server which delivers content. A middle content processing and generation level application server (Rails, Django, Flask). Then a backend database or data store, made up of data sets and database management systems software, which provides access to the data.

Peer to Peer

No special machines that provide a service or manage the network resources. Everyone is considered equal in doing the task. Uses include file sharing, instant and voice communication.

Examples: Skype, Napster

Sources/Further Reading