Given the cost of premium capacity, BI teams will want to follow practices to ensure that these resources are actually required and not being used inefficiently. For example, with large import mode datasets, a simple design change such as the removal of unused columns from a fact table can significantly reduce the size of the dataset and thus the amount of memory needed. By following a series of recommended practices in terms of both modeling and report design, less premium capacity resources will be required to deliver the same query performance and scale.
With small-scale self-service BI datasets and reports, performance tuning and optimization is usually not necessary. Nonetheless, as these models and reports can later take on greater scale and importance, a basic review of the solution can be applied before the content is assigned to premium capacity. For example, the BI/IT team can identify a few small changes to be implemented prior to assigning the pro user's workspace to premium capacity.
The following two sections identify several of the top data modeling and report design practices to efficiently utilize hardware resources.