About a limitation/question on mapd-core



I am a graduate student doing research on database system implementation, and have looked into MapD for couple of months.

While I was looking into the detail of MapD, specifically mapd-core, I realized that MapD’s query execution plan is naive, which is simply the lexicographic order as stated in FROM clause, which cannot guarantee the best performance.

Therefore, I have implemented a cost-based optimizer for MapD to generate optimal query execution plan, and from the experimental results, I found that this optimizer can boost the performance of MapD in some cases.

I would like to share my idea and talk more about this, so please feel free to contact me if you want further information.

Thank you.


We would love to collaborate with you on any MapD core enhancements. Please make a pull request to the mapd core repository and we will review your changes.