In every area of data management, there is a need to collect, process, and return data from temporary objects. Traditional temporary tables and table variables are stored in TempDB and come with the same performance limitations that disk-based tables do.

Using Memory-Optimized Table Variables allows for the creation, management, and disposal of temporary data completely within memory, without the need to store data on disk. This can reduce activity in TempDB while also speeding up processes that operate heavily on, or in-between temporary objects. Combining this feature with In-Memory OLTP and Natively Compiled Stored Procedures can allow for processes to operate completely in memory.

In this lightning talk, Memory-Optimized Table Variables will be introduced, discussed, and demonstrated. The most common use-cases for this feature will be introduced, as well as how they can allow you to overcome some of the limitations found in Natively Compiled Stored Procedures.