Cloud service provider Zunicore is set to launch a beta trial of its hosted private cloud service using graphic processing units (GPUs) for high performance applications.
The service will allow organisations looking to host high performance computing (HPC) applications, such as CGI rendering, protein mapping, derivative market trading or oil and gas reservoir modelling, to specify hardware requirements for a hosted private cloud and provision infrastructure within 15 minutes.
Among the options customers can specify are the number of GPU cores, the time needed to run HPC applications as well as memory and storage. Billing is on-demand, fully automated and charged hourly or monthly. The servers that form the hosted private cloud are interconnected with private switches on secured networks running at 10Gbit/s.
Virtual machine can be provisioned with Linux, Windows and FreeBSD images and can be added to a resource pool within the private network. CPU and Ram can be added dynamically to absorb bursts of traffic to VMs.
While Zunicore doesn’t offer clustering capabilities as yet, Greg Rusu, general manager of Zunicore said that customers would be free to install clustering capabilities on the hosted private cloud themselves and would be taking a “wait and see approach” to see what customers were using in terms of clustering software before deciding which technology to support.
Rusu also said that the cloud service offers PaaS capabilities via PaaS offering StandingCloud. He said that other PaaS might well be supported in the future if the company felt they were appropriate to running HPC applications. “It fits a very good space in the PaaS area as far as high performance computing is concerned”.
The service is still currently in beta at the moment but Rusu said that the full service should launch in July.
