VM High-Performance Computing for Undergraduate Engineering Projects

It is intended that this abstract would be the full paper at 2019 ASEE PSW conference. High performance cluster computing (or simply a super-computing) are essential for undergraduate engineering projects, as they make possible very large simulations which can save both money and time, and help students understand things that cannot be easily replicated experimentally. However, it can cost upwards of a hundred million dollars and require large amounts of space and energy to maintain. Even smaller variations of these systems, though powerful and much cheaper than the larger counterparts, can cost thousands of dollars and again require a lot of space, energy, and time to maintain. For these reasons, normal high performance computing systems can be near inaccessible to undergraduate students. The current project aims to see the effectiveness of using virtualization technology, in place of traditional high performance computing. Virtual machines are a type of software that allow users to run multiple operating systems on the same machine; this means that tasks that cannot be accomplished due to incompatibility with the operating system can still be accomplished on the same machine. The great advantage to virtual machines is the fact that these processes can all be run in the background, allowing users to continue to use the machine for other tasks. In the current project, Windows virtual machines were used as virtual nodes for a new small-scale cluster computing system developed at the Advanced Computing and Simulations Lab of Embry-Riddle Aeronautical University. This new system, utilizing computers already present within the lab as hosts, was tested for its ability to handle moderate computational fluid dynamic simulations, with promising results. Using this virtual cluster setup allows undergraduate engineering major students to gain access to high performance computing at a fraction of the cost normally associated with traditional super-computing.