Non-volatile random-access memory1 has gained recent attention because of its guaranteed data persistence and low data access latency. In-memory key-value stores generally operate by storing log files, which generate disk I/O to prevent data loss from unexpected system failure. As the performance of in-memory key-value stores is bound by disk speed, the advent of NVRAM can be a viable solution to alleviate performance degradation. However, leveraging NVRAM to store entire data is nascent in terms of the cost per capacity. We propose a novel hybrid key-value scheme that consists of NVRAM and dynamic random-access memory, which supports a higher level of data persistence while maintaining high performance. Results from our proposal scheme show outstanding results against NoSQL benchmarks in terms of performance per data persistency. In addition, our scheme provides scalability allowing NVRAM and DRAM to be used without possibility of data loss.
|Title of host publication||Proceedings of the ACM Symposium on Applied Computing|
|Publisher||Association for Computing Machinery|
|Number of pages||4|
|Publication status||Published - 2019|
|Event||34th Annual ACM Symposium on Applied Computing, SAC 2019 - Limassol, Cyprus|
Duration: 2019 Apr 8 → 2019 Apr 12
|Name||Proceedings of the ACM Symposium on Applied Computing|
|Conference||34th Annual ACM Symposium on Applied Computing, SAC 2019|
|Period||19/4/8 → 19/4/12|
Bibliographical noteFunding Information:
This research was supported by the MSIT (Ministry of Science and ICT), Korea, under the SW Starlab support program (IITP-2017-0-00477) supervised by the IITP (Institute for Information & communications Technology Promotion).
All Science Journal Classification (ASJC) codes