April 22, 2014, by Graham Kendall
High Performance Computing Cluster (HPCC)
This post was contributed by Muhammad Yaasiin Oozeer
Computers have become an essential and extensively used tool in scientific research. It is no news to anyone that our home personal computer (PC) itself is quite a small piece of magic which facilitates many of our daily tasks. The same PC can actually be used to run simple scientific tasks and calculations. The benefits of the PC and its capabilities cannot be neglected. Unfortunately research sometimes involves complicated and larger scale calculations which the PC cannot cater for due to its computational limitations. One PC might not do the job. Add in one more and you might just increase your chances. Increase that number and you get closer to getting the job done with a system of connected computers that support each other. This forms the basis for the High Performance Computing Cluster (HPCC).
A computing cluster usually consists of a set of computers which are connected together through local area networks (LAN). Each computer serves as a server and is called a node. Even if every node its own instance of an operating system, usually Linux based, the nodes work in such a way that they may be considered as a single system. Clusters provide more computing power and reliability than single computers and have become a vital requirement in various businesses. Besides their high computing power, one might argue that clusters might be cumbersome as compared to a single computer of comparable speed and performance. It turns out that clusters are much more cost-effective than the later, a factor that cannot be ignored.
How does the HPCC relate to us? Well, research is a key component of our institution and we need to provide the right tools to ensure that it is effectively carried out. So far our 64-node HPCC has performed relatively well in assisting in our research needs but unfortunately its high maintenance requirements and computational limitations are becoming more flagrant. Thankfully we work in close collaboration with the UK campus, and have offered to extend the usage of their HPCC, a staggering 256-node machine, to UNMC. We have warmly welcomed this offer and testing on our side will hopefully start as soon as this week. With such computing possibilities in hand, we have no doubt that research will bloom even brighter at UNMC.