Abstract
Through the massive use of mobile devices, data clouds, and the rise of Internet of Things, enormous amount of data has been generated and analyzed for the benefit of society. NoSQL Databases and specially key-value stores become the backbone in managing these large amounts of data. Most of key-value stores ignore transactions due to their effect on degrading key-value store's performance. Meanwhile, programmable switches with the software-defined networks and the Programming Protocol-Independent Packet Processor (P4) lead to a programmable network where in-network computation can help accelerating the performance of applications. In this paper, we proposed a networking support for transaction processing in distributed key-value stores. Our system leverages the programmable switch to act as a transaction coordinator. Using a variation of the time stamp ordering concurrency control approach, the programmable switch can decide to proceed in transaction processing or abort the transaction directly from the network. Our experimental results on an initial prototype show that our proposed approach, while supporting transactions, improves the throughput by up to 4X and reduces the latency by 35% when compared to the existing architectures.
Original language | English (US) |
---|---|
Title of host publication | Proceedings - IEEE 7th International Conference on Big Data Computing Service and Applications, BigDataService 2021 |
Publisher | Institute of Electrical and Electronics Engineers Inc. |
Pages | 41-49 |
Number of pages | 9 |
ISBN (Electronic) | 9781665434836 |
DOIs | |
State | Published - 2021 |
Event | 7th IEEE International Conference on Big Data Computing Service and Applications, BigDataService 2021 - Virtual, Online, United Kingdom Duration: Aug 23 2021 → Aug 26 2021 |
Publication series
Name | Proceedings - IEEE 7th International Conference on Big Data Computing Service and Applications, BigDataService 2021 |
---|
Conference
Conference | 7th IEEE International Conference on Big Data Computing Service and Applications, BigDataService 2021 |
---|---|
Country/Territory | United Kingdom |
City | Virtual, Online |
Period | 8/23/21 → 8/26/21 |
Bibliographical note
Funding Information:This work was partially supported by NSF I/UCRC Center Research in Intelligent Storage and the following NSF awards 1439662, and 1812537.
Publisher Copyright:
© 2021 IEEE.