It certainly might seem odd to propose an investment to transform the technical computing infrastructure digital manufacturing companies utilize at such a difficult time. Indeed, the industry has been hit especially hard by the global recession that in turn has forced many companies to lay off thousands of employees. However, this could be the most important time to invest in enhancing research capabilities and IT infrastructure. Doing so can help companies to rebound by boosting its design capabilities, increasing productivity and finding ways to reduce costs, or to meet the budget cut requirements imposed by the reality of tough times.
To begin, let’s identify the ways to create the most business value that in turn lead to the biggest returns. Think how a lumberjack would gain by swinging his axe at the biggest tree. This process will be different in some respects for small engineering firms versus the larger more established organizations. Nonetheless, most engineering companies’ largest expense is people and then the IT costs which includes the expensive applications and the infrastructure, power and cooling that support those applications.
Let’s consider ways to improve utilization of the key cost drivers: The engineers and the applications.
The technical applications that a typical engineering firm utilizes are capable of running on workstations at small firms or on a “compute grid” which taps the computing power of multiple machines at larger firms.
The challenge: Run applications faster so they can more efficiently complete simulations which will ultimately give a more competitive edge to the company in the market.
Design engineers generate almost endless demand for compute cycles to run their simulations and optimizations. A typical set of simulations can run from one minute to 60 hours with some jobs taking even longer. For engineers, submitting jobs can become tedious and repetitious when using workstations that are not connected, so finding simpler ways to get work done saves time and creates happier engineers.
Creating a shared environment by lashing computers together into a “grid” allows applications to run on unused compute resources. A compute grid requires special software that schedules and coordinates jobs across all machines on the grid, simply enough called a “scheduler”. A leading example is Univa Grid Engine. Tying computers together can simplify many tasks for an engineer which saves time, a basic lesson learned in the early 90’s when LANs (local area networks) were nascent and “SneakerNet” was everywhere. Imagine the pain without a network that allows us to share files and information. Not sharing a pool of computers is similar when engineers try to run jobs today.
Prepare for the future: Build a Private Cloud
Instead of simply pursuing a software or hardware upgrade or seeking price reductions, map a strategy to transform the entire technical computing infrastructure by considering it at a holistic level. A key example of this is to create a “private cloud.” Infrastructure would be transformed to more efficiently run the same applications as with your previous architecture, but with a focus on reducing systems management overhead. In addition, be sure to architect a private cloud in a way that will support the future capability of reaching beyond your own servers to a “public cloud,” where one could buy additional processing power as needed. This capability is generally called “cloud burst” and is possible only with a scheduler and integrated cloud product capable of supporting and understanding how to dispatch workload in a hybrid cloud.
As cloud-based technologies become more mainstream, companies should look for ways to leverage the cloud within their technical computing infrastructures in order to reduce overall computing costs and improve efficiency – and also to invest in solutions flexible enough to burst into the public cloud when the time is right. The most common perception of "cloud computing" today is the public cloud, where users outsource compute processing and pay only for what they use. But many organizations are finding that it makes sense, prior to investing in a public cloud solution, to simplify and optimize internal processes by employing cloud technology internally.
Extract Value from Suppliers’ innovation
One of the most critical selections that must be made when implementing a private compute cloud for technical applications is the microprocessor architecture, because it directly enables both the cloud’s capabilities and its efficiency. Choosing an architecture for the cloud is somewhat analogous to an airline’s selection of jet engines: It affects the routes they can fly, the fuel consumption, the number of passengers they can carry, etc. It’s a strategic choice.
The microprocessor also must deliver a marked improvement over previous generations because the new computers must provide a sustainable advantage over the life of the equipment—often as long as 4-5 years.
In terms of power consumption, Intel® estimates that for every dollar spent on typical server hardware, 50 cents is spent on power and cooling. So Intel incorporated new power technologies into its latest series of Intel Xeon processors platform to manage power consumption beyond what is already built into the processor. It’s not unreasonable to expect deploying new servers to reduce power consumption, cooling requirements and carbon footprint, often by as much as 65%.
"Investing in new infrastructure routinely, like refreshing older servers, is critical. At Intel IT, we found that continually enhancing our existing infrastructure actually reduces the amount required to maintain and run our environment and therefore is a high-priority investment for us.” —Diane Bryant, CIO Intel Corp.
When considering the holistic strategy be sure to include optimization of one of the most expensive costs and tie allocation of expensive application licenses to the grid. With innovation by Univa, it is now possible to schedule a license to an application with a job. Univa License Orchestrator prioritizes the sharing of limited and expensive application license features according to business objectives by incorporating availability into Univa Grid Engine scheduling decisions.
If you are able to maximize the use of licenses you gain leverage over the cost curve. Fully optimized licenses ensure not only that you get what you pay for, but also that you are armed with facts when considering how to spend your license budget.
More efficient workflow and better usability of the new or improved infrastructure will save time and allow engineers to devote additional time bandwidth to developing innovative solutions for customers. In addition, for companies that move from stand-alone workstations to grid infrastructure, simplified submission of even complex jobs will save more time and increase utilization.
Moreover, by leveraging grid or cloud-enabled infrastructure provided by Univa, engineers can view a daily report on disc usage, CPU usage, software license utilization and more. With visibility of which resources go unused, engineers can make better use of available resources for other simulations, adding greater value for the business.
Creating business value can happen in many ways. In this discussion we talked about how to lower costs while continuing to deliver a high level of performance. To maximize value creation one should consider the challenge of limited availability of capital expenditure (CapEx) funds, so the infrastructure changes have to be funded out of an existing operating budget.
Even while wracked by the global recession, considering an investment project for infrastructure transformation can clearly yield positive results, both hard-dollar savings and softer benefits. Each organization should calculate its own hard savings since mileage will vary. The soft benefits are no less valuable although more difficult to calculate. The end goal is higher quality parts or materials as the case may be and more productive engineers. Other productivity benefits will come in the form of avoided downtime (greater reliability) and one-stop software support offered by Univa for the internal cloud. In addition, with visibility of resources in the private cloud, engineers can make better choices about maximizing their use of those resources.
With excerpts from “Private Cloud Case Study: Corus Automotive Engineering Group (AEG) 14 Ways to Create Business Value With a Univa-Intel® Infrastructure Transformation” authored by The FactPoint Group (www.factpoint.com), a boutique market research and consulting firm based in Silicon Valley, on behalf of Univa. To read the full case study please visit http://www.univa.com/resources/case-studies/cloud-roi-study
To learn how to transform your infrastructure contact Univa today. http://www.univa.com/products?utm_source=dmr_is&utm_medium=sponsor&utm_campaign=transform