Accelerating External Sorting via On-the-fly Data Merge in Active SSDs

Abstract

The concept of active SSDs (solid state drives) has been introduced in order to cope with the demands required to process the ever-increasing volumes of data. In active SSDs, some of the data-processing tasks are offloaded to SSDs, freeing host system resources and improving overall performance of data analysis.

In this paper, we propose a novel active SSD architec- ture focused on improving the external sorting algorithm that is used extensively in data-intensive computing. By performing merge operations on-the-fly in active SSDs, our method can remove the extra data transfer and en- hance the lifetime of SSDs. Our evaluation results on a real SSD platform indicate that the proposed scheme out- performs the traditional external sorting by up to 39%.