Cost Effective Cluster & Cloud Computing

We all have heard the term cluster. It is used to define interconnected computers working together to accomplish big task like very large scientific computing operation. Single high computing machine could be used for the same but owing to the high cost of single machine, interconnected computers are preferred instead.

A computer cluster consists of a group of loosely or tightly connected computer systems that work together so that they can be viewed as a single unified system. Unlike grid computer system, in cluster each computer system is set to perform the same task controlled and scheduled by software running at every connected computer system.

Most big organizations working in technology domain especially upgrade their computers in 8-10 years period, as after that computers become obsolete not withstanding the current technology requirements of the time. The discarded machines are often rendered useless but they could be used to design a unified high speed computing cluster. The discarded machines connected over a high speed Local Area Network can be utilized to form a high computing resource which can be managed by a dedicated task manager computer system and this architecture is known as Beowulf Cluster.

Beowulf Clusters are cost effective clusters. These are created using scrapped and off the shelf components. Although they don't give top-notch performance, their performance is far better than a single computer. A variant of Beowulf Clusters allows OS such as Ubuntu to run on every node and still allow parallel processing. The nodes communicate with each other and perform the computing tasks accordingly.

On the other hand, if we take a look at the cloud we observe that cloud is also having the same underlying concept but with a different terminology or nomenclature. In a way cloud can also be seen as a cluster where the computing can be hired on need basis. It has many different service models depending upon the requirements

For example Software as a Service (SaaS) model involves a cloud service provider to install and maintain software in the cloud and users run the software from their cloud clients over the Internet (or Intranet). The client machine does not require installation of any application specific software. The cloud applications run on the server inside the cloud. SaaS is scalable and system administration may load the applications on several servers. Before cloud the scenario was that every customer would purchase and install their own copy of the application software on their personal computers. Instead with SaaS customer can access application without installing software locally on their machines.

Other examples of cloud are Development as a Service (DaaS) which is a web based and community shared development tool. This is equivalent to locally installed development tools as in conventional approach; Platform as a Service (PaaS) is cloud computing service which provides users with application platforms and databases as a service. This is equivalent to middleware in the conventional system application platform; Infrastructure as a Service (IaaS) means hiring the physical hardware and going in a virtual way. For example servers, storage and system management everything is available in the cloud. This is equivalent to having the infrastructure and hardware as in the conventional machines. In a nut shell, we hire service over internet to have virtual servers, networks, storage from the cloud service provider. This requires the need of a data center, heating, cooling, and maintaining hardware at the cloud hosting site.

Now if we take our Beowulf approach to a next level, the discarded machines which appear outdated & obsolete could be reused to make a cloud. This approach could be used at institutional and organizational level to make own clouds that could be accessed by different research centers within the organization.

This idea of reuse could be beneficial as it leads to cost saving, reduce the problem of e-waste management to some extent and promote green computing.

By - Er. Alok Kumar Agrawal – Asst. Prof. CSE, Chitkara University H.P.