At Los Alamos National Laboratory, scientists and IT administrators need technology that is reliable -- especially when that technology is designed to maintain the safety of the United States' stockpile of nuclear weapons. A large Linux cluster, being built by Bluffdale, Utah-based Linux Networx Inc., has been charged with this daunting task.
Linux Networx also recently announced the opening of its solutions center in Salt Lake City, where companies can test cluster system configurations to meet their needs and goals.
Here, Kim Clark, vice president of engineering at Linux Networx, offers a series of dos and don'ts for implementing Linux clusters.
Do work with a cluster vendor that has proven cluster expertise in designing and building turnkey solutions. A cluster vendor with industry-recognized success and training programs can mean better system reliability and will provide the user with the tools and skills needed to receive the highest return for investment possible.
Do use mature and comprehensive cluster management tools. Linux clusters can be difficult, especially as systems scale up. New cluster management tools empower administrators over these complex systems, lowering the knowledge barriers to adopting the technology. An organization's administrative costs can be greatly reduced when the appropriate management software and hardware are deployed.
Do receive proper Linux cluster training in order to
Do explore the latest cluster technology and available applications. If you're curious as to what hardware and software would work better for your applications, experiment with different clustering configurations. Evaluate various application options to find the one that works best for you.
Don't be afraid of clusters. This technology is quickly becoming the high-performance computing (HPC) solution of choice and, if you don't adopt this technology, your competition will. Linux clusters are proven to speed production and increase organizations' return on investment, and the sooner you implement this technology, the sooner you will see results.
Don't design a system without considering the cooling and power requirements for the cluster and the entire lab area in which the cluster is going to be placed. The correct cooling and power is essential for running a reliable, healthy cluster.
Don't build or design a cluster by purchasing the cheapest components and building it yourself. Not only is this an inefficient method that reduces the reliability of the machine, but this will also create many problems down the road when problems arise and you don't have the necessary service or support.
Don't underestimate the cost to your organization when cluster systems are down. Make sure you work with a vendor that produces high-quality cluster products and delivers high numbers for mean time between failures.
FOR MORE INFORMATION:
FEEDBACK: What advantages has a Linux-based cluster brought to your enterprise?
Send your feedback to the SearchEnterpriseLinux.com news team