Improving performance of cloud key-value storage using flushing optimization
- Authors
- Son, Y.; Kang, H.; Han, H.; Yeom, H.Y.
- Issue Date
- Sep-2016
- Publisher
- Institute of Electrical and Electronics Engineers Inc.
- Citation
- Proceedings - IEEE 1st International Workshops on Foundations and Applications of Self-Systems, FAS-W 2016, pp 42 - 47
- Pages
- 6
- Journal Title
- Proceedings - IEEE 1st International Workshops on Foundations and Applications of Self-Systems, FAS-W 2016
- Start Page
- 42
- End Page
- 47
- URI
- https://scholarworks.bwise.kr/cau/handle/2019.sw.cau/60835
- DOI
- 10.1109/FAS-W.2016.22
- ISSN
- 0000-0000
- Abstract
- Key-value store is an essential component with an increasing demand in many scale-out environments, including social networks, online retail, and cloud services. Hence, modern key-value storage engines provide many features, including transaction, versioning, replication, etc. In storage engines, transaction processing provides atomicity and durability by using write-Ahead logging (WAL), which flushes log data before the data page is written to persistent storage in synchronous commit. However, according to our observation, WAL is a performance bottleneck on key-value storage engine since the flushing of log data to persistent storage incurs significant overhead of lock contention and fsync() calls even with various optimizations in the existing scheme. In this paper, we propose an approach for improving performance of key-value storage by optimizing the existing flushing scheme combined with group commit and consolidate array. Our scheme aggregates multiple flushing of log data into a large request on the fly and completes the request early. This scheme is an efficient group commit that reduces the number of frequent lock acquisition and fsync() calls in the synchronous commit while supporting same transaction level that the existing scheme provides. We implemented our scheme on WiredTiger storage engine and evaluated that our scheme improves the performance by 1.3-5.5x on the key-value workload compared to the existing scheme.
- Files in This Item
- There are no files associated with this item.
- Appears in
Collections - College of Software > School of Computer Science and Engineering > 1. Journal Articles
![qrcode](https://api.qrserver.com/v1/create-qr-code/?size=55x55&data=https://scholarworks.bwise.kr/cau/handle/2019.sw.cau/60835)
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.