The Completion Suggester can be often be memory intensive. Is there any way to setup a ES cluster so that only certain nodes hold data for the completion suggester?
Ideally, I could have a few nodes that solely handle completion queries and the data nodes would solely handle data.