We use Dataiku in AWS with Instances (Design, Deployer, Automation). The first three month all instances where running 24/7 in the cloud. To reduce our cost we have created managed tags on each instance called start_time and stop_time.
Setup: In the cloud (AWS Lambda) we created a function that reads the current time and checks if an instance is running. The function starts or stops the instance if the current time is between start_time and stop_time. A job executes the function every 5 min.
We were able to reduce out AWS costs by 50% compared to the first 3 month. Our design node does now run Monday - Friday 7:00 to 21:00, for Deployer and Automation node we have set different start and stop times. I believe that every customer could use this kind of feature in their cloud, if it was available through fleet manager.