An equi-join algorithm based on low-update conditions

Solid State Drives (SSD) adopt flash memory as their data storage medium, so their random reading speed is hundreds of times faster than traditional magnetic disks. However, comparison with magnetic disks, traditional query methods can´t bring about the high-speed in the case of flash-memory. The query algorithms of some existing flash-based DBMS can be only applied to stable and simple data environment, which has server limitations. When a small amount of data is changed, query is low efficiency. This paper presents a new equi-join algorithm between two tables, which optimizes join query of flash memory DBMS in low-update conditions. This algorithm saves the cost of I/O according to PAX-based layouts, and reduces intermediate results generated from join so as to decreases the expensive write cost of flash memory. When data sets are not frequently updated, it can obtain good capability by using join indexes. Experiments show that the algorithm combines join index with PAX-based layouts effectively, tremendously reducing the read-write operations of flash-memory and improving query efficiency.