In our experience most data sets loaded into analysis services cubes will benefit from InMemory loading for several reasons.
- Query performance will be significantly improved as there will no longer be a reliance on disk IO to deliver the data. Instead, it will go straight to memory, which is significantly faster.
- Combining data from across several dimensions is faster.
- There is no need to cross analyze data files from the file system (to find non-empty data, like in cubes).
- Data can be joined directly InMemory (similar to a relational database).
Thus, the TARGIT InMemory Database can be used to solve the performance bottlenecks common with traditional analysis services cubes.
Additionally, a set of BI functions have been added to the engine to replace calculations we are currently executing in MDX. For example, the balance calculation for Finance Balance and Inventory Balance can be made using the built-in balance function in the TARGIT InMemory Database. This is because it is a built-in function and data is sitting in memory, the performance is significantly better than the MDX calculated members done today.