Evaluating the Impact of AES-256 Encryption on Network Performance: An Analysis of Transfer Time, Latency and Throughput

 Background and objectives - The file size of plain text impedes network performance. This level of impairment should be used as the benchmark for evaluation of encryption algorithms. The objective of this study is primarily to compare the effects of plain text and AES-256 encryption on network performance. Additionally, it will determine the effect of the encryption on plain text data size and the relationship between data size and selected network performance metrics. Materials and methods- Plain text and AES-256 encrypted text of various data sizes were categorized into two groups- small to medium data size (10-100MB) and very small data size (0.1-0.9MB). The percentage increase in file size caused by encryption was recorded for each data. They were serially transmitted through EVENG simulated network environment. Transfer time, latency, and throughput were determined. The results were further evaluated using comparison of means, Pearson and Spearman’s correlations, line graphs and scatter plots. Results - The increases in file size after encryption varies from 0.031% at 0.1MB to 0.00003% at 100MB. Graph lines of the metrics against data size are predominantly coincident but differ in pattern between the two categories. Scatter plots and correlation coefficients show a significant (p<0.05) positive correlation between transfer time and data size of each text in both categories, latency in the 10-100MB but not in the 0.1-0.9MB categories (p>0.05), or throughput in general. Curiously, a significant (p<0.05) strong positive correlation exists between throughput and data size in the 0.1-0.5MB range. A throughput saturation value starts at 60 MB for plain text and at 30 MB for cipher text. Conclusion- AES-256 encrypted text and plain text have similar effects on network performance probably because of the former’s negligible effect on data size. Data size positively correlates with transfer time. Similar correlation exists with latency only in the 10-100MB range but not in the 0.1-0.9 MB range and not with throughput. An early, but, transient correlation occurs between data size and throughput in the 0.1-0.9MB range.