June 18, 2007
When Altair Engineering announced the new licensing model attached to the release of its PBS Professional 9.0 grid management software, it took a big step toward the future that could (and, many would argue, should) spark an evolution in the way users pay for their grid middleware.
Targeting users who want a truly on-demand HPC environment, the new PBS Pro licensing model allows them to enable entire datacenters, as well as desktops and other disparate computing resources, with the software but only pay for the number of licenses (per core, virtual or physical) they want to run concurrently. What this means, at a very basic level, is that someone running a dual-socket, dual-core SMP machine could enable all four cores with PBS Pro, but only license three, for example, to run the software at any given time. But this is much more to it than this.
Michael Humphrey, vice president of PBS GridWorks for Altair, likens the changing of the HPC landscape to the switch from horse-drawn buggies to automobiles, citing advances in grid computing, virtualization, multi-core processors and on-demand computing as catalysts. However, he notes, this evolution must overcome barriers, especially in terms of inflexible and costly licensing models, if the software industry is to complement the hardware industry, which already has adjusted prices to ease the burden on users. “Lots of software companies are still selling wagon wheels,” said Humphrey, “when we should be changing our model and selling things that are more modern and up-to-date.”
To address these concerns, Altair has made cost and flexibility priorities. From a price perspective, the company has slashed by about 75 percent the per-license cost of PBS Pro. Users of PBS Pro 9.0 will pay only $13.50 per concurrent license, which comes out to only $1,728 per year to license 128 cores in a model where an old Cray machine costs as much to license as a Pentium 1-powered desktop. Although the price per license certainly is bound to attract attention, it is along the flexibility front where Altair really looks to be separating itself from the pack.
Using patented technology based on Macrovision’s FLEXlm licensing model, licenses of PBS Pro reside in a central license server until they are drawn out by jobs needing more power. When the job completes, the licenses go back into the server and are available for other jobs that might need them. Should users wish to lock PBS Pro licenses onto specific hardware so, for example, a high-powered cluster always has licenses available to run jobs, that also can be arranged -- and, if need be, those locked licenses can be pulled back without disrupting service; they simply will return to the license pool when the jobs finishes.
On top of this, Altair allows users to mesh its definition of “on-demand” with that of resource providers, such as HP or IBM, which provide computing on-demand. The ability to utilize PBS licenses outside the organizational firewall allows Altair customers to take advantage of vendors offering high-performance computing resources on-demand, and customers even have the option of “locking in” licenses on these external CPUs, as well. However, it’s not just the customers who win in this scenario, as the ease of moving licenses to external resources helps to alleviate one of the biggest barriers to outsourcing computing jobs: working with software providers to find a mutually agreeable pricepoint for the additional licensing. Says Altair’s Humphrey, “With Altair, not only do you not have to come talk to us and negotiate something, you can just do it automatically.”
And because PBS Pro 9.0 doesn’t differentiate between different types of processing units -- whether they be virtual machines, single-core processors or individual cores on a multi-core chip -- Altair believes its software will help solve the issue of how license in multi-core environments to help customers make the best use of their hardware investments. For AMD, who is banking on the success of its multi-core processors, this definitely is good news. David Rich, AMD’s director of high-performance computing, believes that Altair’s model will help facilitate positive experiences for organizations that make the switch to multi-core, which is good news for the chipmaker. “We have hardware, so we want people to discover that it’s useful to run things faster, for example,” he said. “If there’s a huge licensing or a huge economic step to go through before you even try to run it on twice as much hardware … we don’t like that.”
“Grids are really important,” says Rich, and previous per-core licensing models have been detrimental to end-users, who often end up paying huge licensing fees when they decide to upgrade their resources. This isn’t necessarily fair, said Rich, as chipmakers are now maximizing speed by increasing the number of cores instead of increasing the speed of a single core, for which users did not have to pay more in licensing costs. He cites as an example AMD’s compute grid, which currently boasts between 7,000 and 8,000 servers, none of which are single-core. With PBS Pro, they could enable every single server without being forced to buy upfront licenses for each core -- a number that could range between 14,000 and 32,000.
Of course, issues around multi-core processors -- both how to license and how to optimize applications for multi-core environments -- have been picking up steam lately, and the reason is evident. “For the most part, whether you think you want it or not,” said Rich, “when you buy a server in the future, you’re going to get more than one core per socket.”
For William Fellows, principal analyst with The 451 Group, whose Grid Adoption Research Service has been tracking the grid market for three years, Altair’s new licensing model certainly will have an impact on the company, if not the grid software market as a whole, but he’s not positive it will provide the blueprint for licensing in the still-evolving worlds of virtualization and multi-core processors. That said, Fellows says licensing has been a key barrier to the adoption of grid technologies in almost all markets, with the notable exception of financial services (due to the deep pockets and referential leverage possessed by companies in this space), and he believes other grid vendors could stand to benefit from adopting similar licensing models.
Fellows also notes that more flexible grid middleware licensing models don’t necessarily correlate with more flexible licensing from other vertical market-specific ISVs, but he sees the potential for advance in certain areas, such as electronic design automation, due to demand from industry bodies. For its part, Altair, said Humphrey, is considering licensing its patented technology to ISVs in the oil and gas and life sciences markets, as well as horizontal vendors like The MathWorks and Wolfram Research, with whom it has pre-existing relationships.
When it’s all said and done, though, the real criterion for gauging how likely we are to see other vendors moving to similar models is how it affects the bottom line at Altair. Interestingly, Humphrey acknowledges that while the move could be risky from a revenue perspective, the company actually expects to see an increase in revenue. For one, he said, Altair has reconstructed contracts with some partners to streamline business processes, and the company also could see users of its free, open source Open PBS make the switch to PBS Pro now that the licensing costs have been so drastically reduced. Additionally, says Humphrey, the company has gained confidence from a similar move with its HyperWorks suite of products nine years ago, when revenue actually ended up increasing.
And although competitors in the grid space might consider replicating to various degrees Altair’s new business model, Altair believes it has one thing going for it that its rivals do not: another highly successful line of business. Humphrey sees Platform and its LSF product as his main competition, but believes that a company like Platform might have to undergo some serious changes in order to hedge against the possibility of serious revenue hits. Altair’s GridWorks division, on the other hand, can fall back on the profitability of the company’s HyperWorks line of business, which holds top positions in the automotive, aerospace and other manufacturing-focused markets. “We’ve got the advantage that we can make this aggressive move and, in the worst case scenario, if it is cannibalistic toward our revenues,” said Humphrey, “we can weather through that because the big ship, HyperWorks, can keep us going in the right direction.”
May 16, 2013 |
When it comes to cloud, long distances mean unacceptably high latencies. Researchers from the University of Bonn in Germany examined those latency issues of doing CFD modeling in the cloud by utilizing a common CFD and its utilization in HPC instance types including both CPU and GPU cores of Amazon EC2.
May 10, 2013 |
Australian visual effects company, Animal Logic, is considering a move to the public cloud.
May 10, 2013 |
Program provides cash awards up to $10,000 for the best open-source end-user applications deployed on 100G network.
May 08, 2013 |
For engineers looking to leverage high-performance computing, the accessibility of a cloud-based approach is a powerful draw, but there are costs that may not be readily apparent.
05/10/2013 | Cleversafe, Cray, DDN, NetApp, & Panasas | From Wall Street to Hollywood, drug discovery to homeland security, companies and organizations of all sizes and stripes are coming face to face with the challenges – and opportunities – afforded by Big Data. Before anyone can utilize these extraordinary data repositories, however, they must first harness and manage their data stores, and do so utilizing technologies that underscore affordability, security, and scalability.
04/02/2012 | AMD | Developers today are just beginning to explore the potential of heterogeneous computing, but the potential for this new paradigm is huge. This brief article reviews how the technology might impact a range of application development areas, including client experiences and cloud-based data management. As platforms like OpenCL continue to evolve, the benefits of heterogeneous computing will become even more accessible. Use this quick article to jump-start your own thinking on heterogeneous computing.