We are using a split stack and there is a predictable load pattern in the query group.
Currently we are using a predictive auto-scaling policy but maybe we could also benefit from spot instances for the query group?
