Return to search

Performance benchmarking of data-at-rest encryption in relational databases

This thesis is based on measuring how Relational Database Management Systems utilizing data-at-rest encryption with varying AES key lengths impact the performance in terms of transaction throughput of operations through the process of a controlled experiment. By measuring the effect through a series of load tests followed by statistical analysis, the impact of adopting a specific data-at-rest encryption algorithm could be displayed. The results gathered from this experiment were measured regarding the average transactional throughput of SQL operations. An OLTP workload in the benchmarking tool HammerDB was used to generate a transactional workload. This, in turn, was used to perform load tests on SQL databases encrypted with different AES-key lengths. The data gathered from these tests then underwent statistical analysis to either keep or reject the stated hypotheses. The statistical analysis performed on the different versions of the AES-algorithm showed no significant difference in terms of transaction throughput concerning the results gathered from the load tests on MariaDB. However, statistically, significant differences are proven to exist when running the same tests on MySQL. These results answered our research question, "Is there a significant difference in transaction throughput between the AES-128, AES-192, and AES-256 algorithms used to encrypt data-at-rest in MySQL and MariaDB?". The conclusion is that the statistical evidence suggests a significant difference in transactional throughput between AES algorithms in MySQL but not in MariaDB. This conclusion led us to investigate further transactional database performance between MySQL and MariaDB, where a specific type of transaction is measured to determine if there was a difference in performance between the databases themselves using the same encryption algorithm. The statistical evidence confirmed that MariaDB vastly outperformed MySQL in transactional throughput.

Identiferoai:union.ndltd.org:UPSALLA1/oai:DiVA.org:his-21309
Date January 2022
CreatorsIstifan, Stewart, Makovac, Mattias
PublisherHögskolan i Skövde, Institutionen för informationsteknologi
Source SetsDiVA Archive at Upsalla University
LanguageEnglish
Detected LanguageEnglish
TypeStudent thesis, info:eu-repo/semantics/bachelorThesis, text
Formatapplication/pdf
Rightsinfo:eu-repo/semantics/openAccess

Page generated in 0.0019 seconds