Sangfor HCI is deeply integrated with GPUs, providing an alternative to traditional GPU demand scenarios. It allows customers to quickly build GPU cloud platforms and offers features such as GPU resource quota management, user self-service, and billing. Through extensive customization and development, HCI effectively manages the full lifecycle of GPU resources, significantly improves management efficiency, and supports diverse business operations in various scenarios.

Highlights of the GPU scenario solution based on Sangfor HCI:

1. GPU Resource Pooling

In vGPU mode, GPU resources across the entire cluster can be flexibly scheduled as a unified resource pool. License and hardware resources are automatically released to the resource pool when users shut down.

2. GPU Resource Visualization

Administrators can easily monitor the GPU resource utilization, available GPU resources, partitioning of each GPU, and VM usage through the control panel.

Full-Lifecycle Management of GPU/vGPU Resources

The solution supports full-lifecycle management of GPU/vGPU resources, including resource application, switching, monitoring, and deallocation. All operations can be performed through the HCI web UI.

3. Hybrid Cluster in HA Mode

The solution allows the combination of common and GPU servers to form a cluster, providing HA rules for VMs on the clustered servers to ensure business continuity.