Cluster (computers)

Computer cluster. Set of computer equipment that behave like a single Supercomputer . They are mainly used for the solution of high computational cost problems related to science, engineering and commerce.

Summary

[ hide ]

  • 1 Story
  • 2 Development
  • 3 Classification
    • 1 Stages
  • 4 Combinations
  • 5 See Also
  • 6 Source

History

Cluster technology began in the 1950s and 1960s, in the midst of a technological revolution.

The main foundation that gave rise to the existence of clusters is Amdahl’s Law, which describes mathematically how much can be expected to do a series of tasks in parallel using an architecture that allows it. This law is applicable to any means of multiprocessing, be it in hardware (that is, machines with multiple CPUs ) or network environments (computer clusters).

Consequently, the history of the first groups of computers is more or less directly linked to the history of the principles of networks, as one of the main motivations for the development of a network to link computing resources , in fact the creation of a cluster of computers. Networks were conceptually invented by the RAND corporation in 1962 .

Using the concept of a packet switched network, the ARPANET project managed to create in 1969 what was possibly the first basic computer network, based on the cluster of computers by four types of computer centers. The further development of ARPANET is what is known as a global network or INTERNET .

The first commercial cluster-type product was ARCnet , developed in 1977 by Datapoint but was not commercially successful and the clusters were not successful until VAXcluster produced the VAX / VMS operating system in 1984 . The idea was to provide the benefits of parallel processing, while maintaining data reliability and unique character. VAXcluster, VMScluster is still available on HP OpenVMS systems running on Itanium and Alpha systems.

Two other notable cluster business principles were the Tandem Himalaya (circa 1994 , with high availability products) and the IBM S / 390 Parallel Sysplex (also circa 1994, primarily for company use).

The history of computer clusters would be incomplete without pointing out the critical role played by the development of the Parallel Virtual Machine ( PVM ) software .

This open source software based on TCP / IP communications allowed the creation of a virtual supercomputer – an HPC cluster – performed from any of the connected TCP / IP systems.

Freely heterogeneous clusters have been the pinnacle of this model, rapidly increasing FLOPS globally and far exceeding the availability of even the most expensive supercomputers.

PVM and the use of low-cost PCs and networks led, in 1993 , to a NASA project to build cluster supercomputers.

In 1995 , the invention of the beowulf – a cluster style – a computer farm designed according to a core network product with the specific goal of “being a supercomputer” capable of performing strong and parallel HPC calculations.

This spurred the independent development of Grid computing as an entity, even though the Grid style revolved around that of the Unix operating system and the Arpanet.

Development

The clusters have been developed in various branches, such as applications for high computing solutions, Web servers and electronic commerce , as well as high performance databases, among other uses.

The increase in the use of clusters is due to the easy access nowadays to inexpensive computers with powerful microprocessors , the existence of high-speed networks, as well as the development of software to carry out parallel programming or distributed computing .

But mainly the existence of clusters is motivated by the need for computational power to solve very complex problems.

Classification

Computer clusters can be classified in the combination of the following general characteristics:

  1. High performance
  2. High availability
  3. Load balancing
  4. Scalability

Stages

Two stages can be described for the assembly of a cluster:

  • HardwareSelection : it is where the computers that form our cluster are chosen. These can be

homogeneous (similar hardware) or heterogeneous (different hardware).

  • SoftwareSelection : this is where it is defined that the Operating System or Middleware will be used by the computers.

Combinations

Combinations of all are possible. We can find a computing cluster where there is hardware homogeneity but software heterogeneity. In this situation, generally, a homogeneous middleware is present from the software side.

We can also find heterogeneity of hardware with the same software. The situations depend on the political and / or economic contexts of the responsible institutions or people.

 

Leave a Comment