It is often said that one of the key capabilities of the Cloud is the delivery of services to businesses that otherwise would not be able to afford them – the capital investment required is just too big. There isn’t a sector of computer usage where this is more true than with High Performance Computing (HPC), which makes the announcement by US computer vendor, SGI, of its Cyclone HPC Cloud services of particular interest.
SGI is one of those companies that has contenders in the Supercomputer Top 500 list – these are the machines used by big universities and research establishments for a wide range of bleeding edge research and development work. It owns Cray Computers, which currently tops the Top 500 list with its Jaguar machine, which is located at the US Department of Energy’s Oak Ridge Leadership Computing Facility. This has posted a 1.75 petaflop/sec. performance speed and has a current theoretical peak capability of 2.3 petaflop/sec. This is achieved using nearly a quarter of a million processor cores. Just for completeness, one petaflop/sec. refers to one quadrillion calculations per second.
These are truly machines where it can be said that if you need to ask the price you certainly cannot afford them. Yet there are many businesses that could make use of such a capability if they could access the power and software, which is where a Cloud-based delivery model comes into play.
Cyclone is based on SGI’s Altix server systems, which come in a variety of cluster configurations that allow systems to be scaled up and scaled out. They can run Linux from either Novell SUSE or Red Hat and are available either in SaaS, where users run the applications SGI provides, or IaaS (Infrastructure as a Service) where SGI hosts the customers’ own applications.
Either way, as SGI’s CEO, Mark Barrenechea, observed, “users will pay for what they use rather than worrying about complex computers or infrastructure.”
To begin with, at least, the applications software available in the SaaS service is geared towards the science and engineering communities, with commercial packages such as OpenFOAM, NUMECA, Gamess and Gromacs. These are likely to attract the interest of any business or organisation where fast solutions to complex engineering issues such as computational fluid dynamics are important, and where budget restraints demand cost-effectiveness.
But the systems technology is also of direct interest to businesses and organisations where any form of complex risk analysis and planning are important, so it can be expected that the availability of applications in these area will start to appear in due course for the SaaS.
[1]