Cray To Bring Supercomputing Services to Microsoft Azure
If existing high-performance computing (HPC) isn't enough for you, Microsoft is bringing the supercomputing capabilities provided by Cray to its Azure public cloud.
This is a noteworthy deal because Cray has been regarded for decades as the leading provider of supercomputing systems. Cray's supercomputers are capable of processing some of the most complex, high-performance and scientific workloads performed. The two companies today announced what Cray described as an "exclusive strategic alliance" aimed at bringing supercomputing capability to enterprises.
While the term "exclusive" is nebulous these days, this pact calls for the two companies to work together with customers to offer dedicated Cray supercomputers running in Azure datacenters for such workloads as AI, analytics and complex modeling and simulation "at unprecedented scale, seamlessly connected to the Azure cloud," according to Cray's announcement.
The deal marks the first time Cray is bringing its supercomputers to a cloud service provider, according to a statement by Peter Ungaro, the company's president and CEO. The two companies will offer the Cray XC and Cray CS supercomputers with its ClusterStor storage systems for dedicated customer provisioning in Azure datacenters and offered as Azure services.
"Dedicated Cray supercomputers in Azure not only give customers all of the breadth of features and services from the leader in enterprise cloud, but also the advantages of running a wide array of workloads on a true supercomputer, the ability to scale applications to unprecedented levels, and the performance and capabilities previously only found in the largest on-premise supercomputing centers," according to Ungaro.
Cray's systems will integrate with Azure Virtual Machines, Azure Data Lake storage, Microsoft's AI platform and Azure Machine Learning (ML) services. The Cray Urika-XC analytics software suite and Microsoft's the CycleCloud orchestration service that Microsoft now offers following its August acquisition of Cycle Computing, can be used for hybrid HPC management.
The fact that Microsoft would want to bring Cray into the Azure equation is not surprising given CEO Satya Nadella's focus on bringing supercomputer performance to the company's cloud, a priority he demonstrated last year at the Ignite conference in Atlanta. In last year's Ignite keynote, Nadella revealed some of the supercomputing functions Microsoft had quietly built into Azure including an extensive investment in field programmable gate arrays (FPGAs) throughout the Azure network backbone, bringing 25Gbps backbone connectivity, up from 10Gbps, combined with GPU nodes.
Nadella stepped it up a notch in his opening keynote at the recent Ignite gathering held last month in Orlando, where he revealed extensive research and development effort focused on trying to one day offer quantum computing. The ability to offer this form of high-performance computing requires breakthroughs in physics, mathematics and software programming that are still many years away from achievement. While IBM and others have long showcased some of their R&D efforts, Microsoft' revealed it too has had been working on quantum computing for many years.
Nadella and his team of researchers said Microsoft will release some free tools by year's end that will let individuals experiment with quantum computing concepts and programming models. The pact with Cray will bring supercomputing processing capabilities to Azure that will solve the most complex challenges in climate modeling, precision medicine, energy, manufacturing and other scientific research, according to Jason Zander, Microsoft's corporate VP for Azure.
"Microsoft and Cray are working together to bring customers the right combination of extreme performance, scalability, and elasticity," Zander stated in a blog post. While it's not immediately clear to what extent, if any, Cray will be working with Microsoft on quantum computing, it's a safe bet that they will do so at some level, if not now, then in the future.
Posted by Jeffrey Schwartz on 10/23/2017 at 12:12 PM