Research On SSD-based Join Query Optimization Technology In Array Database

Massive data processing is a heated topic currently and an increasing number of researches attach importance to it.In the domain of scientific research,complexity analysis queries lead to much consumption of the network and disk IO.It is due to the scientific data analysis operation contains a large number of Join queries,and the Join queries are CPU and IO intensive.Join queries are even much less effective in the largescale data set which results to low efficiency of scientific data analysis.Even array databases developed specifically for scientific data are often not effective in cope with the dilemma.Join queries in scientific applications are IO intensive and CPU intensive queries,which cost a lot of resource.SSD can speed up Join queries by virtue of its fast read and write speed,and ultimately improve the efficiency of scientific data analysis.In this paper,the join queries on massive data in distributed environment are optimized combined with the characteristics of astronomical data.The main contents include(1)the Target Join algorithm is proposed to reduce the overhead of the array database join queries.It outperforms 0.164 to 2.67 times than queries in SciDB.(2)the JAPJA algorithm is proposed to decrease the IO cost of the array database join queries.It outperforms 0.18 to 5.27 times compared to SciDB.(3)The JAPJA algorithm based on SSD(JAPJABS)is proposed to reduce random access on mechanical hard disk.The results demonstrate that JAPJABS is improved by 0.02 to 2.67 times on the basis of JAPJA algorithm and 0.95 to 6.66 times compared to SciDB.