Thinking Outside the In-Memory Box [AD-307]

Speaker: Dmitri Korotkevitch

Duration: 75 minutes

Even though in-memory OLTP can significantly improve performance of OLTP systems, it comes with an extensive set of limitations that often make refactoring of existing systems cost-ineffective and prevents widespread adoption of the technology. However, in-memory OLTP can be beneficial in other common use cases besides replacing on-disk with memory-optimized tables. 

This session discusses several of these use-cases:  
* Improving performance of ETL processes with memory-optimized tables
* Using the in-memory OLTP engine as a replacement for the distributed cache and session state server
* Using natively compiled procedures to optimize imperative code and calculations
* Enhancing performance of batch import operations with memory-optimized TVPs

Finally, this session shows several techniques that help address in-memory OLTP limitations by utilizing horizontal and vertical partitioning and combining the data from on-disk and memory-optimized tables.