Author(s): Raghu Gopal
Cloud computing, the business of renting computer power and other technology infrastructure, has become a standard and convenient service allowing enterprises to easily outsource parts of their IT needs. Nonetheless, the concept has its limitations, particularly when it comes to dealing with complex modelling and simulations.
This week, Microsoft announced a partnership with supercomputer manufacturer Cray to bring its devices and storage system to Microsoft's Azure cloud platform. According to Peter Ungaro, president and CEO of Cray, the arrival of his company's high-end systems on Microsoft's cloud service makes supercomputing attainable to more businesses that were previously unable to purchase or maintain an on-premises Cray system.
The agreement will provide more high-performance computing power to Azure customers, especially those that manage massive sets of data used for artificial intelligence. Supercomputers from Seattle-based Cray are widely used by government agencies and private businesses to handle heavy-duty computing needs. The partnership means Azure and Cray customers can keep all their data in one place, and harness the power of both systems. Businesses that subscribe to Cray's service will have access to a dedicated, customized supercomputer that connects directly into an Azure data centre.
The supercomputing-as-a-service offering will be hosted in select Azure data centres. The companies claim that resource provisioning on Cray's XC and CS product series, along with the accompanying ClusterStor storage systems, will be done according to each customer's needs.
Availability of Cray supercomputer resources on the Azure platform will enable researchers, analysts and other professionals to perform tasks like training artificial intelligence deep learning models in areas such as medical imaging and autonomous vehicles, performing whole genome sequencing thus shortening the time from computation to cure, conducting crash simulations and carrying out computational fluid dynamic modelling. The service will allow workloads that would typically need massive investments in hardware and IT management to be shifted to virtual machines in the cloud. The companies say that Cray supercomputers will be able to run some tasks in days, and in some cases just minutes, instead of months or weeks on lesser hardware.
For Microsoft, the move is the latest in its efforts to bulk up its cloud-based high-performance computing offering and lure businesses, scientific research labs and other organisations with resource-intensive workloads to Azure. Since Cray's focus remains squarely on the high-performance computing needs of researchers in both academia and industry, the partnership could offer its users a lower-cost option as the need for more high-powered computing spreads.