Many organizations are now finding they need to run applications that require IT infrastructure historically associated with high-performance computing (HPC) environments. The challenge they face is that most of those organization don’t have access to the capital budget required to build, much less manage, IT at that level of scale. To make HPC more accessible, Verne Global this week announced a HPC-as-a-service offering that can be invoked on demand whenever needed.
Dominic Ward, managing director for Verne Global, says the hpcDIRECT service is hosted in a data center based in Iceland that provides access to bare-metal servers based on Intel Xeon processors. Those systems are then networked together using Infiniband and Ethernet networks based on switches provided by Mellanox Technologies.
In addition to new classes of applications, Ward says Verne Global expects science and engineering organizations to employ hpcDIRECT as extensions of their existing environments as well as a facility for running standalone applications.
Increased usage of CAD/CAM applications in manufacturing and rising interest in analytics being applied to massive amounts of Big Data are two of the more compelling new use cases requiring HPC infrastructure. But Ward says the challenge many organizations face is that the cost of acquiring HPC infrastructure as a capital expense is prohibitive. The hpcDIRECT service provides a means of treating HPC as an operational expense that can be deployed much faster than any internal IT organization is likely to similarly provision on their own, says Ward.
“It really speeds up setting up an instance of HPC,” says Ward.
Verne Global is not the only hosting or cloud service provider to offer access to a HPC service. But as interest in HPC-level applications expands, it is clear that the number of services options that organizations can now avail themselves of is also starting to expand.