- Difficulty
- Intermediate
- Time
- 45–60 min
State the promises
Describe the two workload classes by latency need and work shape, and give each a service-level objective you are willing to be judged on.
What to demonstrate: Are the promises concrete enough to arbitrate a conflict later?
Admit and batch
Define what is admitted when the devices are full, and what happens to a long running batch when a latency-sensitive request arrives.
What to demonstrate: Does admission protect the promise that matters at that moment?
Trade fairness against preemption
Explain whether the throughput class may be preempted, what it costs when that happens, and how you would keep one class from starving the other over an hour.
What to demonstrate: Can the candidate quantify the cost of the policy it chose?
Lose a GPU
One device leaves the pool while both classes have work in flight. Describe memory placement, what happens to that work, and how the lost capacity is accounted for in each promise.
What to demonstrate: Does the recovery path keep the promise accounting honest?