Microsoft and Cray are to link supercomputers directly into Azure data centres, providing high-end scalability for AI workloads
Microsoft is to offer Cray supercomputing power via its Azure cloud platform, in a deal aimed at those looking to speed up the training of artificial intelligence (AI) systems, carrying out large-scale analytics processes or other processing-intensive chores.
Separately, Microsoft launched an update to Windows 10 with a feature aimed at stemming data losses caused by ransomware.
The Cray deal follows Microsoft’s acquisition of high-performance computing (HPC) firm Cycle Computing in August.
It will see Cray XC and Cray CS supercomputers with attached Cray ClusterStor storage systems made available for customer-specific provisioning in select Azure data centres.
The companies didn’t specify the geographic locations planned for the systems.
The supercomputers are to be integrated with Azure virtual machines, data storage and Microsoft’s cloud AI platform and machine learning services, said Microsoft corporate vice president Jason Zander.
“All of this is provided in the cloud with the most datacentres worldwide,” he wrote in a blog post.
Cray chief executive Peter Ungaro said the deal gives Azure users access to the benefits of supercomputing, including scalability and and a level of performance previously only found in large supercomputing centres.
HPC and supercomputing technologies are in greater demand as companies increasingly require large amounts of processing capacity for AI and analytics workloads, as well as others including creating detailed models and simulations.
Cray and Microsoft said the supercomputers could be used by researchers, analysts and scientists in fields such as medical imaging, drug discovery and the development of autonomous vehicles, where AI plays a central role.
In the pharmaceutical and biotechnology fields supercomputing can shorten the time it takes to carry out whole genome sequencing, while automotive and aerospace product engineers can use the platform to conduct crash or fluid dynamic simulations.
Energy companies can speed up oilfield analysis and reduce exploration risks by improving seismic imaging fidelity, the companies said.
In May, Azure competitor Amazon Web Services (AWS) said it would collaborate with Nvidia to offer high-performance systems for AI training based on Nvidia GPUs. Google cut a similar deal with Nvidia in February.
Also this week Microsoft outlined a new feature in the recently launched Fall Creators Update to Windows 10 (version 1709) that aims at protecting users’ data from ransomware, a type of encryption malware that has become increasingly disruptive in recent months.
The “Controlled folder access” feature in Windows Defender, the platform’s set of built-in security tools, protects important folders from being modified by applications that aren’t cleared by a whitelist.
When switched on, the feature by default protects common folders where important data is kept, such as the Documents folder, but users can determine which folders should be protected and which applications are granted access.
Early testers said the feature successfully blocked common ransomware variants such as Locky from encrypting the data held in protected folders. The feature doesn’t disable ransomware, however, meaning everything outside of the list of protected directories would still be lost in a ransomware attack.
Controlled folder access can be enabled in Windows Defender Security Centre or using Group Policy, PowerShell or mobile device management systems.
How well do you know the cloud? Try our quiz!