SQ27
A GKE cluster runs both GPU-intensive ML training and standard API workloads. How should you isolate these workload types?
- ARun everything on one default node pool
- BUse separate node pools with taints/tolerations and appropriate machine types
- CUse separate GKE clusters per container
- DOnly use CPU for all workloads
Show answerHide answer
Answer: B
Separate node pools allow GPU-accelerated machines for ML and standard machines for APIs. Taints and tolerations ensure workloads are scheduled to the correct pool.