Cluster Computing & Architecture

Cluster Computing & Architecture

Introduction

Cluster computing is referred to a set of tightly or loosely connected computers that operate together in many various compliments in a way that they can be viewed on a single server . In cluster computing the computer components are connected to each other with the help of LAN (Local Area Network) with every individual computer as an individual node used by the server that runs its own instance on the server (Baker et.al, 2009, pp. 87-125).

Cluster computing emerged as an outcome of convergence of numerous trends of computing including the availability high speed networks, low cost microprocessors & software for enhanced performance disseminated computing (Fuerte, 2009).

Computer cluster is a group of computers that are interconnected and work together in a coordinated fashion, so that in many respects they can be considered as a single. Cluster computers usually interconnected through a local area network. Clusters of computers are commonly used to improve speed and / or reliability of a single computer by providing cost-benefit ratio is better than a single machine in terms of performance or reliability (Fuerte, 2009).

History of Cluster Computing

As in many other areas, even here the need was the mother of invention. Customers are those who conceived the idea of cluster computers as soon as they could not get enough speed outputs from a single, or as required for backup. Need for clusters appeared in the 60s of the 20th century or even the 50s (Barak & La'adan, 2008, pp. 361-372).

Foundation engineering construction of computer clusters and parallel system was assumed by the chief architect of IBM, Jin Mammadli (Gene Amdahl). In 1967 published a position paper Mammadli parallel computing, known as the Law Mammadli. Mammadli Law describes mathematically the acceleration performance that can be expected through the admissions process compared execute them sequentially (Baker et.al, 2009, pp. 87-125).

The article defined the engineering basis for long computing - processors and cluster computers. The main difference between the two types of technologies these is whether the relationship between the - processors (interprocessor) done within the computer or between computers over the network. For this reason, the history of the first computers clustered closely linked to the history of the network computer communication. One of the strongest motivations for creating and developing these networks would link resources, thus creating a cluster of computers de - facto (Barak & La'adan, 2008, pp. 361-372).

Networks packet switching developed by the RAND Corporation in 1962. Project ARPANet use the idea of packet switching, which in 1969 created the first network where computers were connected to four different centers (within each of them was' cluster computers).

Project ARPANet expanded and became what we call the Internet today. The Internet is, if all clusters of computers, because it unifies all computing resources into logical resource centers. Network ARPANet established paradigm prevalent in all existing computing clusters: the use of packet switching networks for communication - process performed between the processors in different locations (Fuerte, ...