High-performance teaching requirements
Kwantlen Polytechnic University’s Entertainment Arts faculty needed a practical way to support graphics-intensive applications and 3D systems for students and faculty. The cost and operating effort of an on-premises GPU VDI environment were central constraints.
Cost, access and capacity
The design had to consider application experience and the cost of providing GPU sessions. Moving to a single cloud without evaluating capacity and redundancy would not by itself establish an acceptable operating model.
The architecture in context
Conceptual architecture illustrating decision boundaries. It does not represent a client environment.
The work delivered
- Current-state analysis of graphics-intensive teaching requirements.
- Solution design across suitable AWS, Azure and Google Cloud options.
- Scalability, redundancy and cost analysis.
- A proof of concept to evaluate the target approach.
Automation and alternative capacity
The project narrative describes GPU compute accessed through virtual desktops, with Azure considered as an alternative platform when AWS capacity allocation was delayed. Terraform and Ansible supported repeatable infrastructure configuration.
The architecture’s useful lesson is to evaluate capacity, access and operating effort together. A secondary provider introduces its own identity, data and support dependencies that need validation.
Value of the approach
- A clearer basis for evaluating cloud GPU desktop cost and capacity.
- Repeatable provisioning and configuration practices.
- An architecture intended to improve student/faculty access to demanding applications.
- A tested starting point for operational and redundancy decisions.
Applying the lesson
Use representative teaching applications, user devices and network conditions in a proof of concept. Agree what a successful session looks like and how data, support and capacity will be managed during the term.
Questions and answers
Which teaching environment did the project address?
The source case describes Kwantlen Polytechnic University’s Entertainment Arts faculty and its graphics-intensive applications and 3D systems.
#Should another institution expect the same cost outcome?
No fixed saving is assumed. Application licenses, session patterns, cloud capacity, data access and redundancy choices determine the economics of another environment.
#Why start with a proof of concept?
It provides evidence about application experience and operating constraints before a broader rollout. That evidence can change both the architecture and the cost assumptions.
#