Further developing an already successful partnership after their joint launch of vGPU virtual desktop solutions based on KVM, NVIDIA has officially designated Sangfor as one of the first vendors in Asia to support their groundbreaking vGPU9.0 virtualization solutions. Cooperation between Sangfor and NVIDIA continues to make great strides in the AI data center, machine learning and scientific computing areas.
Continuous Innovation and Deep AI Integration
In June 2019, NVIDIA released version 9.0 of their GPU virtualization solution, offering a number of optimization features for various computing scenarios, as well as a new vGPU license-vComputeServer (vCS). With a primary focus on the optimization and expanded use of artificial intelligence, machine learning, and scientific computing, vCS makes it easier for users to manage GPU-accelerated virtualization servers, while maintaining existing workloads and reducing overall operating costs. VCS, powered by four NVIDIA V100 GPUs, improves deep learning speed 50 times, compared to CPU-only servers, bringing performance even closer to direct computing level.
Source: NVIDIA Official Blog
NVIDIA vGPU solutions offer a total of four types of vGPU. The new vCS license in vGPU9.0 is a further refinement of their previously offered support for AI and HPC scenarios.
Sangfor Cloud Platform--Compatible with vCS Licenses
The widespread implementation of AI and deep learning have become more and more important to technology stacks. Thus, those responsible for IT infrastructure construction pay an increasing amount of attention to the process of business critical application operation and management in complex AI and big data scenarios.
Based on the newly developed vCS license in NVIDIA vGPU9.0, Sangfor's hyperconverged team, through deep cooperation with NVIDIA, implemented functions like GPU resource quota management, user-generated services and billing on their hyperconverged management platform, aCloud. In addition, aCloud provides full life-cycle management of GPU resources through in-depth, customized research and development. These developments make it possible to continue delivering a simple, reliable, and secure private cloud environment to users, while HCI provides much needed stability, as it hosts a variety of business critical applications making use of AI and deep learning.
Core Functions of the GPU Cloud Platform based on HCI
• Life cycle management of GPU resources
• Visualized operation of GPU resources
• Unified authentication management based on LDAP
• Multi-tenant, self-service, metering and billing
(hyperconverged infrastructure) helps users build their GPU cloud platform quickly, and greatly improves efficiency of management. Sangfor HCI
provides high-level performance and high-density GPU virtualization, proven especially useful in the construction of AI GPU cloud platforms in Universities.
The strategic partnership between Sangfor and NVIDIA has enabled both enterprises to provide delivery compatibility and continuously optimized solutions from the endpoint to the cloud. Faced with increasingly fierce competition, Sangfor cloud computing has always upheld the principles of compatibility and openness, and actively promotes industry innovation and integration. In the future, Sangfor plans to continue to work with NVIDIA to help AI fusion industrialize solutions and build an enterprise service ecosystem.
About Sangfor Technologies
Sangfor’s cloud computing sector builds the simplest, most stable, and highest-performing hyper converged architecture, offering an overall solution that includes private cloud, hybrid cloud, and managed cloud. Sangfor is committed to providing comprehensive cloud computing solutions and services, from local to cloud and from private to hybrid cloud, so as to build a solid foundation for the digital transformation of government, enterprise and public institutions. Sangfor also helps users effectively lower cost, shorten application on-line time, reduce complexity of operation and maintenance, enable productivity and focus on business innovation.