Case studies

 15 Case Studies for Classroom Group Discussion (Grid Computing)

 1. High‑Performance Medical Imaging Grid

A hospital shares machines with GPUs for MRI image processing. Only certain machines have expensive licensed medical software.
Discussion: How should the grid scheduler assign jobs? What policies protect sensitive medical data?

 2. Student Scavenging Grid at a University

Student lab computers run grid jobs only when idle.
Discussion: How does scavenging ensure computers remain usable? What happens to jobs when students log in?

 3. Weather Prediction Data Grid

A climate research team uses data striping across many storage nodes to speed up huge dataset access.
Discussion: How does striping help? What are the risks if one storage node fails?

 4. Multi‑Cluster Campus Grid

Several labs form clusters; clusters combine into one hierarchical grid.
Discussion: How does distributed grid management ensure scalability? What issues occur when scheduling across clusters?

 5. Software Licensing Bottlenecks

Only a few machines contain a costly simulation software.
Discussion: How should the scheduler handle limited-license jobs? Should jobs be queued differently?

 6. Company‑Wide CPU Donation Program

Employees donate unused CPU cycles from their desktops.
Discussion: What role does donor software play? How can the grid prevent interfering with employees’ active work?

 7. Remote Sensor Data Aggregation

A project collects data from hundreds of field sensors; storage nodes vary in capacity.
Discussion: Which storage resources should the grid prioritize? How can communication delays be reduced?

8. Research Lab Reserved Resources

A university reserves certain machines exclusively for physics projects.
Discussion: How do resource reservation policies affect fairness? Should unused reserved resources be temporarily shared?

 9. Distributed Rendering for a Film Studio

A film studio uses a grid to render animation frames; some machines have better GPUs.
Discussion: How can the scheduler match jobs to hardware capabilities? Should GPU-powered nodes get higher-priority tasks?

10. Financial Institution Grid with Heavy Security Requirements

A bank wants to submit risk‑analysis jobs but machines differ in policies and security levels.
Discussion: How should the grid handle nodes with strict security? Can jobs be restricted to specific architectures?

11. Mixed Operating System Environment

The grid includes Windows, Linux, and macOS nodes with different file systems.
Discussion: How do platform differences complicate job scheduling? How does the grid choose compatible nodes?

12. Scientific Experiment Requiring Synchronized Subjobs

An application splits itself into many communicating subjobs using MPI.
Discussion: What happens if network speed varies between nodes? How can the grid ensure the subjobs stay synchronized?

13. A Grid Facing Network Congestion

High‑data applications overload the grid’s network links.
Discussion: How should communication resources be managed? Should large data transfer jobs be scheduled during low‑traffic periods?

14. Real‑Time Disaster Response Grid

Emergency services deploy a grid to process real‑time data during a natural disaster.
Some nodes have high CPU speed but low storage; others are opposite.
Discussion: How can the grid match jobs to the right mix of CPU and storage resources? What scheduling strategy works best?

 15. Grid Failure and Autonomic Recovery

A grid’s management node fails during peak workload.
The system supports autonomic (“self‑healing”) recovery.
Discussion: What recovery steps should autonomic management take? How does redundancy improve resilience?