For large-scale deployments of Power BI in which multiple types of datasets and workloads will be supported (import refreshes, as well as DirectQuery and Live connection queries), BI teams can consider multiple gateway clusters. In this approach, each gateway cluster is tailored to meet the specific resource needs (RAM and CPU) of the different workloads, such as large nightly refreshes or high volumes of concurrent queries in the early mornings.
For example, one gateway cluster could be composed of two gateway instances with a relatively high amount of available RAM on each gateway server. This cluster would have resources available during the most intensive scheduled refresh operations (for example, 4 A.M. to 6 A.M.) and would be exclusively used by import mode Power BI datasets and any Azure Analysis Services models that also regularly import data from...
