DRAT - A Dynamic Resource Allocation Tool for Estimating Compute Power in a Cybersecurity Engineering Learning Facility

Abstract

Cybersecurity laboratory infrastructure has direct impact on the quality of student learning experiences. Because of this, the computing education field has developed a variety of approaches to designing and implementing these learning facilities. Yet, little work has gone into how to properly size cybersecurity laboratory infrastructure relative to student population and curricular compute power demands. The result has been laboratory infrastructures that do not scale with degree programs. Consequently, laboratories are either underpowered, thus limiting learning experiences, or overpowered which wastes financial resources. Accordingly, this work presents DRAT, an open-source software tool, for estimating necessary compute power in a cybersecurity engineering learning facility. More specifically, DRAT is designed to estimate the required discrete compute power on a per exercise basis in a cybersecurity engineering learning facility operating in a private cloud model. Such discrete estimations are intended to communicate physical host hardware requirements such as physical CPU core count, virtual RAM, and total Hard Disk space. The first step in designing DRAT was to forge a model estimator function. Then, we identified a series of scalar abstractions representing learning facility hardware infrastructure and behaving as conversion factors between the model function and output. Because the goal of this work was to provide estimates for cloud compute power requirements, DRAT outputs the number of physical cores, total RAM, total Disk, and total (virtual or physical) Network interfaces required to run the indicated scenario. The implication is that such estimates can inform purchasing and configuration decisions which directly impact student learning outcomes.