TY - GEN
T1 - An I/O isolation scheme for key-value store on multiple solid-state drives
AU - Kim, Hwajung
AU - Yeom, Heon Young
AU - Son, Yongseok
N1 - Publisher Copyright:
© 2019 IEEE.
PY - 2019/6
Y1 - 2019/6
N2 - High-performance storage devices, such as Non-Volatile Memory express Solid-State Drives (NVMe SSDs), have widely adopted in data centers. Especially, multiple storage devices improve higher I/O performance compared with a single device. However, the performance can be reduced in the case of workloads with mixed read and write requests (e.g., key-value stores) even though multiple storage devices are adopted. The reason is that read requests can be blocked until the processing for write requests is finished. In this paper, we propose an I/O isolation scheme to improve the performance of key-value store for multiple SSDs. In our scheme, we separate read and write operations for multiple files. For example, we classify generated files of key-value store and perform read/write operations in the separated storage devices according to the characteristics of each file. To do this, we deploy files by considering their characteristics on multiple SSDs. We implement our scheme in RocksDB and evaluate its performance with six NVMe SSDs compare with RAID-0 configuration. In the experimental results, we demonstrate that our scheme improves performance compared with an existing scheme (i.e., RAID-0).
AB - High-performance storage devices, such as Non-Volatile Memory express Solid-State Drives (NVMe SSDs), have widely adopted in data centers. Especially, multiple storage devices improve higher I/O performance compared with a single device. However, the performance can be reduced in the case of workloads with mixed read and write requests (e.g., key-value stores) even though multiple storage devices are adopted. The reason is that read requests can be blocked until the processing for write requests is finished. In this paper, we propose an I/O isolation scheme to improve the performance of key-value store for multiple SSDs. In our scheme, we separate read and write operations for multiple files. For example, we classify generated files of key-value store and perform read/write operations in the separated storage devices according to the characteristics of each file. To do this, we deploy files by considering their characteristics on multiple SSDs. We implement our scheme in RocksDB and evaluate its performance with six NVMe SSDs compare with RAID-0 configuration. In the experimental results, we demonstrate that our scheme improves performance compared with an existing scheme (i.e., RAID-0).
KW - High-performance storage device
KW - Key-Value Store
KW - LSM-tree
UR - http://www.scopus.com/inward/record.url?scp=85071419118&partnerID=8YFLogxK
U2 - 10.1109/FAS-W.2019.00049
DO - 10.1109/FAS-W.2019.00049
M3 - Conference contribution
AN - SCOPUS:85071419118
T3 - Proceedings - 2019 IEEE 4th International Workshops on Foundations and Applications of Self* Systems, FAS*W 2019
SP - 170
EP - 175
BT - Proceedings - 2019 IEEE 4th International Workshops on Foundations and Applications of Self* Systems, FAS*W 2019
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 4th IEEE International Workshops on Foundations and Applications of Self* Systems, FAS*W 2019
Y2 - 16 June 2019 through 20 June 2019
ER -