Southeast University

World-class Universities
ZStack Cloud assisted Southeast University in building two types of clusters:Cluster One: GPU servers are used in high-load HPC scenarios, combined with the ZStack Cloud bare metal module for faculty use.Cluster Two: Primarily used for low-load HPC teaching scenarios, intended for student use.After the new cloud platform was established, the cloud hosts originally deployed on another vendor’s cloud platform were migrated to ZStack Cloud.

Business Background

Faced with the need for unified management and migration of cloud platforms, Southeast University encountered the following business challenges:

  1. The need to provide unified management capabilities for both cloud platforms and physical machines.
  2. Support for GPU passthrough or  vGPU capabilities.
  3. The ability to migrate virtual machines deployed on a competitor’s cloud platform to the ZStack Cloud platform via V2V.
  4. Provision of multi-tenant self-service capabilities.
  5. Utilization of the university’s existing parallel file system storage (GPFS).
  6. Simplified operations and maintenance, with unified management allowing personnel to monitor the cloud platform’s operational status in real-time.

Solution Deployment

After analyzing and confirming Southeast University’s cloud adoption needs, ZStack Cloud assisted in setting up two types of clusters:

  • Cluster One: GPU servers are used in high-load HPC scenarios, combined with ZStack Cloud’s bare metal module for faculty use.
  • Cluster Two: A cluster composed of 2U servers primarily used for low-load HPC teaching scenarios, intended for student use.

The primary storage uses the parallel file system GPFS, with servers connected to the primary storage via an IB network.

Upon completion of the new cloud platform, Southeast University migrated the cloud hosts originally deployed on a competitor’s cloud platform to ZStack Cloud.

Project Benefits

  1. The client achieved unified management of both HPC heavy-load and teaching scenarios.
  2. The client was able to utilize existing GPFS resources, fully protecting their investment.
  3. Post-cloud adoption, unified management of GPU resource passthrough and vGPU was achieved.
  4. Full reuse of existing resources was realized, with unified management of resources.
  5. Unified management of bare metal and cloud platforms, allowing operations on bare metal through the cloud platform’s unified interface, facilitating user maintenance.
  6. Provision of multi-tenant self-service, reducing maintenance difficulty and costs, enabling faculty and students to independently request, create, and release cloud resources for teaching and experimentation.