問題文
A logistics team trains an image model and asks an administrator why their session with a GPU request waited for a new node while ordinary sessions started at once. Which description of the node groups explains the wait?
選択肢
- The workbench keeps the GPU instances awake at all times and the group that holds them answers a request at once.
- The workbench keeps a separate autoscaling group for the GPU instances and that group takes the sessions asking for at least a single GPU device.
- The workbench keeps a separate autoscaling group for the GPU instances and that group also absorbs the plain sessions of a busy CPU group.
- The workbench keeps one autoscaling group for every kind of instance and the scheduler picks the node type that suits a queued pod.