Abstract
External sorting algorithm is an important operation in computer science and technology. External sorting algorithm plays a key role in big data processing and database performance. With more and more hardware resources integrated into SSDs, it makes them full of parallelism. This paper proposed a new external sorting algorithm called FSSort trying to exploiting the internal parallelism to accelerate the sorting process. FSSort can fully parallel the data loading and make the data writing in batchs in the run merge phase. Moreover, it makes the pipeline easier. Therefore, FSSort can make efficient use of the internal parallelism of SSDs. Experiments on unsorted data and partially sorted data on three different type of SSDs are conducted. Experimental results show that for unsorted data, the FSSort algorithm can achieve a speedup of 1.39~1.50 comparing to MONTRES, the previously proposed state art of sorting algorithm.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
