Aggregations are precalculated summary tables that provide query results much faster than if you were to query the full fact table. For example, the sample FactResellerSales
table contains over 60000 rows. Instead of reading through 60000 rows to find the sum of total product cost, SSAS can read a single row in the aggregation table. In real-world applications with billion-row fact tables, the difference is much more pronounced, especially when summarized values are grouped by dozens of attributes. Unfortunately aggregations don't come free of charge; they are calculated as part of partition processing and therefore increase the total processing duration. Aggregations could also use significant amount of disk space. Therefore, designing aggregations is an art of balancing performance improvement with the overhead of processing time and additional storage.
SSDT and SSMS offer two wizards for designing aggregations: Aggregation Design Wizard and Usage Based Optimization...