I/O Performance Improvement of Secure Big Data Analyses with Application Support on SSD Cache
Kenji Nakashima, Joichiro Kon, Saneyasu Yamaguchi · 2018
For utilizing private big data, such as DNA data, encryption and anonymization are essential for preserving privacy. However, encryption and anonymization sometimes increase the size of data largely. Thus, increasing I/O performance for large-scale data is important. Caching data with Solid State Drive (SSD) is a popular method for by using SSD as cache, which is a proposed method for improving access performance of data in Hard Disk Drive (HDD). In this paper, we focus on SSD cache and discuss a method for improving I/O performance of a big data application using SSD cache. First, we evaluate the basic I/O performance with and without SSD cache. Second, we reveal the behavior of flash cache. Third, we propose a method for improving I/O performance of a large scale DNA application with SSD cache. Fourth, we evaluate the proposed method and demonstrate that the method can improve I/O performance effectively.