Paper Accepted at a KIISE-Listed Outstanding Software Conference (First author: Lee Chang-gyu; Corresponding author: Professor Kim Young-jae)

Lee Chang-gyu, Byun Hyun-ki, Noh Sung-hyun, Kang Hyun-gu

(combined master's–doctoral, 4th semester) (master's graduate, February 2019) (master's, 3rd semester) (combined master's–doctoral, 2nd semester)

Graduate students of the Department of Computer Science & Engineering — Lee Chang-gyu (combined master's–doctoral student), Byun Hyun-ki (master's graduate, February 2019, now a researcher at TmaxSoft), Noh Sung-hyun (master's student) and Kang Hyun-gu (combined master's–doctoral student) — presented a paper under the supervision of Professor Kim Young-jae of the Department of Computer Science & Engineering at the 2019 ACM International Systems and Storage Conference (SYSTOR), held at the IBM Haifa Research Lab in Israel from 3 to 5 June 2019.

The ACM International Systems and Storage Conference (SYSTOR) is a distinguished conference on computer systems, covering big data infrastructure, storage and file systems, virtualization technology and more. SYSTOR is also listed as an outstanding conference in the KIISE 'Revised List of Outstanding Software Conferences 2018'.

The accepted paper is titled 'Write Optimization of Log-structured Flash File System for Parallel I/O on Manycore Servers'. It identifies the problem that, on manycore servers with tens or hundreds of CPU cores, file systems cannot draw out the full performance of high-performance SSDs connected over PCIe. As a solution, the paper proposes a technique for accelerating file synchronization for data consistency in existing log-structured file systems using non-volatile memory, a next-generation memory technology, together with a range-based locking scheme allowing multiple CPUs on a manycore server to access a file simultaneously — greatly improving performance.

For the enormous data volumes and high processing performance demanded by big data processing and scientific computing, the number of cores per CPU and per server and the speed of SSDs are increasing rapidly. Applying these research results in practice on such high-performance servers is expected to yield greater data processing efficiency.

• Download the paper

- SYSTOR '19 conference programme (https://www.systor.org/2019/program.html)

- ACM Digital Library (https://dl.acm.org/citation.cfm?id=3325828)

• Download the KIISE Revised List of Outstanding Software Conferences 2018