IO throttling
Atlas should be able to throttle IO in a way that does not allow excessive IO to degrade a node or take a node offline. Current functionality has no defense against a high IO data job other than to allow a node to fail.
1
vote
James
shared this idea