Implement a Storage Engine to Improve Query Performance Based on a Computational Storage Device

Recently, as the importance of database management systems that manage massive amounts of data has been increasing, a computing environment for memory-centric or storage-centric is required rather than processing-centric computing. By implementing a distributed file system separated into computing and storage, large amounts of data can be processed in parallel to be stored and managed efficiently, and availability and scalability can be improved while maintaining multiple copies in storage. However, indiscriminate use of Compute Storage Devices (CSD) resources can lead to lower performance than traditional DataBase Management System (DBMS) and slower query speeds due to storage loads. This paper proposes a CSD-based Storage Engine that can overcome the limitations of traditional DBMS technology and efficiently manage large amounts of data in an environment where the DBMS engine and storage are separated. Scheduling is performed to speed up query execution and data is stored and managed based on cache data to reduce host engine and CSD resource usage.