September 15, 2014
HPC (high-performance computing) clusters can help you do more design optimizations in less time than your engineering workstations. But, you say, HPC is too complex and too costly. Only the big outfits can afford to mess with it. We’re a mid-sized outfit and our IT crew doesn’t need to be fiddling with extending open source code, going batty over workload management, handholding users trying to set up simulations and playing cluster cop when everyone insists their simulation is the highest priority of all. We’ll stick with the workstations. They may slow us down crunching away for hours or days on some job, but that’s how we have to roll.
You were right. Once. Times and technology have changed. Your IT people do not have to cope with those hassles anymore and, for that matter, HPC clustering is also now easier for those who use it. Today’s Check it Out link takes you to a paper that first debunks some common misperceptions about HPC clusters and then provides some straight-talking advice to help guide you as you consider an HPC cluster for your company. It even has a crash course in cluster terminology if you need it, although the paper takes pains to avoid overwhelming you with HPC insider lingo.
“HPC Clusters Drive Design Optimization: Six Steps to Making Clusters a Reality for any Business” is designed to help you determine if now is the time to make the leap to HPC cluster computing and, if so, provide you with guidance to move ahead. It was produced by the editors of Desktop Engineering and sponsored by IBM Platform Computing. This writer had nothing to do with it, but has to admit that it’s a good piece of work. Here’s why.
This 12-page paper kicks off with a brief statement of the pressures surrounding you, such as complex electro-mechanical products, tight budgets, tighter deadlines, more materials and strict regulatory environments. It then segues to its main argument — advanced HPC clustering is within your reach — and it does so on a very important note: “Your powerful multi-core workstations set simulation and design optimization in motion, [but] there are limitations from both a performance and workflow standpoint.”
Modern HPC clusters extend your total productivity by enabling you to perform more and more realistic simulations throughout your design cycles while speeding up your workflow. The IT complications of the early generations of HPC clusters have been addressed. For example, workload management software now eliminates most set-up and deployment challenges. User-defined policies handle aligning compute resources with demands. Browser-based interfaces simplify deploying, using and managing a cluster, and application templates make it easy for end users to submit jobs. In short, administration, training and support costs, cluster downtime and other headaches of the past are gone.
The paper next guides you through the steps to follow to make HPC clustering work for you. These steps range from defining your business needs and project scope to assembling your project team and specifying your cluster through deployment, optimization and maintenance. Each step provides insights, tips and guidance to help you work your way through the project. What’s best here is that each of these discussions recognize that no two companies are alike, but that there are a variety of common issues to be addressed.
The paper has a number of tables as well as sidebars on successful HPC deployments that well illustrate the potential change to the way you go about business that HPC clustering offers. One table, for example, asks you a series of pointed questions to help you determine if you are ready to make the leap.
At the end of the paper, you’ll find a number of hyperlinks to additional materials, including ebooks, videos, webinars and white papers. Perhaps the most intriguing link of all is embedded on page 6 as part of the discussion on defining your business problems and the scope of your HPC project. This link takes you to IBM’s Platform Total Cost of Ownership Calculator. This is a terrific tool that poses just a handful of easily answered questions that can help you determine the potential monetary effects of making the transition to an HPC cluster computing environment.
“HPC Clusters Drive Design Optimization: Six Steps to Making Clusters a Reality for any Business” brims with commonsense. It’s readable by IT admins, finance people and everyday designers and engineers. Anyone considering or advocating for an HPC cluster will do themselves well by reading and considering all that this paper has to offer. Download your complimentary copy from today’s Check it Out link.
Thanks, Pal. – Lockwood
Anthony J. Lockwood
Editor at Large, Desktop Engineering