- 著者
- Kosei Masumura Takashi Hoshino Hideyuki Kawashima
- 出版者
- Information Processing Society of Japan
- 雑誌
- Journal of Information Processing (ISSN:18826652)
- 巻号頁・発行日
- vol.30, pp.552-561, 2022 (Released:2022-08-15)
- 参考文献数
- 50
- 被引用文献数
- 4
We attempted to improve the performance of Silo, a concurrency control protocol for in-memory database management system that does not perform efficiently under high-contention workloads. Adaptive backoff is known as an effective optimization method under high-contention workloads. As a result of analysis, we found that its efficacy lies in the elimination of unknown conflicting events rather than in the reduction of re-conflicting events already known, which has been considered so far. On the basis of this analysis, we propose a new method, the thread activity management method (TAM). We conducted experiments comparing Cicada, the state of the art concurrency control protocol, and Silo with our proposed method. The results demonstrated that the proposed method enabled Silo to significantly outperform Cicada. We found that cache misses are vitally relevant to performance.
言及状況
外部データベース (DOI)
Twitter (1 users, 3 posts, 1 favorites)
Admission control can be more effective when it directly controls concurrency when possible (e.g. at server level), rather than request rate, because excess concurrency->bottleneck resource saturation->throughput collapse. Related DB paper: https://t.co/PKwLcz0Y7z
@pkhuong Thanks for reminding me of that paper. In my case I have completely reliable failure detection, so that’s not a rationale for nonblocking CM. In any case, directly controlling concurrency is likely more effective than backoff: https://t.co/PKwLcz0Y7z
Sometimes I come across a paper that would never be accepted at a top conference but is exactly what I need for my current problem. This one addresses retry + backoff in the context of OCC and finds that directly controlling concurrency is more effective: https://t.co/PKwLcz0Y7z