Security Time Prediction of Big Data Workflows with AES Algorithm-aware Simulation

Abstract

Increasing the number of applications with data-intensive workflows, like healthcare workflow, has led to a tendency to embed highly popular cloud computing in the matter of delivering substantial computing resources, and ensuring the security and performance of these complex applications is of utmost importance while estimating sufficient execution time and the amount of resources required for deployment. While that cloud-based execution time includes CPU time, I/O operations, and security time, particularly for workflows involving big data. However, in our previous work, we introduced a methodology to model, simulate, and predict the runtime of big data workflow, including intricate Next Generation Sequencing (NGS) pipelines. This simulation approach provides a realistic estimate of the runtime for test data that is much larger than the training data used. In this paper, we tackle the problem of predicting big data workflow security time performance using a simulation model that takes the (Advanced Encryption Standard) AES algorithm into account. To precisely assess the runtime impact, our methodology entails modeling and simulating the encryption and decryption procedures within big data workflows. We demonstrate our method's effectiveness in generating precise runtime predictions and validate it using an NGS pipeline implemented in e-Science Central. For ensuring optimization performance without compromising data security, the results show the importance of considering security overhead in the NGS pipeline. However, this work makes contributions to the field by applying a practical simulation framework based on WorkflowSim to predict security-related performance impacts. The results confirm an exponential relationship between the stable execution time of implementing security algorithms and the volume of processed big data, indicating that the time doubles as the data volume doubles.