Impact of TLS Overhead with Forward Secrecy on Segmented Network for Multiple Cloud Database Systems

Abstract

Abstract Cloud database serves flexible, affordable, and versatile database frameworks. Indeed, even the cloud database is secure with Transport Layer Security (TLS). However, the performance overhead that TLS presents while executing procedures on the five major No SQL databases: Mongo DB, Apache Cassandra DB, Amazon Dynamo DB, Redis DB, and Couch DB regarding dormancy. We proposed a Multiple Replica Database Architecture (MRD-ARC) to investigate TLS execution overhead with forward secrecy for five NoSql databases, we performed two tests mimicking average database utilization designs with TLS cipher suite ECDHE-prime256v1. We examined connection pooling, where an application involves a solitary connection for some databases activities. Then, we considered one request for each connection. An application opens a connection, executes an interaction, and quickly shuts the connection in the wake of finishing the activity based on read-only throughput, read-only response, and connection throughput. Our experimental result shows that applications that cannot persevere through tremendous overhead execution should be sent inside an appropriately portioned network instead of empowering TLS with forward secrecy. Applications utilizing TLS should utilize a connection pool instead of a connection for each request. We also compare the TLS overhead with forward secrecy in terms of latency of five databases and found that Mongo DB is the best cloud database system with less overhead.