Cluster Computing, or simply Cluster Computer is a collection of computers connected by a communication network. The purpose of Cluster Computing is to distribute a very complex process between the various components of the cluster computers. In essence, a problem that requires a lot of processing power to be solved is decomposed into separate subproblems which are solved in parallel. This obviously increases the computing power of the system. You can also compare to see the difference of Cluster Computing with Fabric Computing.
Types of Cluster Computing
There are three types of cluster computing system : Fail-over, load balancing and high performance computing, the first two which are probably the most common:
Fail-over Cluster Computing : The machines’ work is continuously monitored and when one of the two host stops working the other machine takes over. The aim is to ensure a continuous service.
Cluster with load balancing : It is a system where work requests are sent to the machine with less load.
HPC Cluster Computing : The computers are configured to provide extremely high performance. The machines break down the processes of a job on multiple machines in order to gain in performance.
Characteristics of Cluster Computing
Cluster Computing system makes multiple computers to act as a single computational resource and the various components of Cluster Computing are resources dedicated to the operation of the whole, then the server cluster is a high performance server, instead of putting the burden on a single system, divides the workload (ie functions, for example – mail servers, web servers, database servers and file servers) on multiple machines. In fact, Cluster Computing is a form of distributed system.
To create a Cluster Computing system one must :
- Have an operating system able to operate computers as a cluster (Like Linux)
- Hardware for high-performance network in Cluster Computing
- Ability of Cluster Computing to run systems in parallel
Advantages of Cluster Computing
The advantages of this Cluster Computing system are:
- Economical : In fact these Cluster Computing systems are up to 15 times cheaper than traditional supercomputers with respect to which, for the same performance.
- Scalability : Since the resources are distributed in case of Cluster Computing.
Easy to upgrade and maintain.
- Reliability : As the Cluster Computing system continues to operate even in case of failures in parts of it, obviously with lower performance.
Disadvantages of Cluster Computing
The main disadvantages of Cluster Computing are:
- Difficult to manage and organize a large number of computers
- Poor performance in the case of non-parallelizable applications
Physical space needed is considerably greater than that of a single server
- Increased power consumption compared to a single server
Usage of cluster computing for high performance
Using Cluster Computing can provide significant performance gain. System X, the supercomputer of Virginia Tech was the twenty-eighth most powerful supercomputer on Earth. It is a cluster from 12.25 TeraFLOPS consists of 1100 Apple XServe G5 dual processor 2.3 GHz ( RAM from 4 GB , HD SATA 80- GB ) running Mac OS X and interconnected via InfiniBand (Information provided by TOP500 Supercomputing Sites.)
This article on Cluster Computing and its Usage is written by Prof. Mahendra Lal Chaturbedi.