Summary
International Conference on Emerging Technologies for Communications
2022
Session Number:S2
Session:
Number:S2-5
Performance of Deeply Analyzing Application Switch
Satoshi Ito, Akihiro NAKAO, Masato OGUCHI, Saneyasu YAMAGUCHI,
pp.-
Publication Date:2022/11/29
Online ISSN:2188-5079
Summary:
Kanaya et al. have proposed a method of caching KVS data in application switches to improve the performance of Cassandra, which is a database management system based on a key-value store, running on a network. In the method, a developer implements a function for analyzing Ethernet frames in an application switch. The switch analyzes the frames, extracts query and response data, and caches the data in the switch to improve the response performance. However, this existing method does not analyze lexically and syntactically frames. It extracts information from the data at fixed addresses in the frame payload. Therefore, variable-length table names, key names, and values can not be analyzed. This issue can be solved by implementing lexical and syntax analyzers in an application switch. Naturally, this will cause an overhead due to the analyses. This paper evaluates the overheads of these analyses on KVS performance.