- What is Beocat?
- What does Beocat Do?
- Who Uses Beocat?
Beocat is Kansas State University's Beowulf computing cluster. A cluster is, on the fundamental level, two or more computers that are used together to solve a problem. This type of computing, as it relates to research at K-State, is called High Performance Computing (HPC).
Beowulf Clusters are scalable performance clusters based on commodity hardware, on a private system network, with open source software and infrastructure. Performance can be improved proportionally by adding machines. The commodity hardware can be any of a number of mass-market, stand-alone compute nodes as simple as two networked computers each running Linux and sharing a file system, or in the case of Beocat, 122 nodes, 1000 CPUs, and 1648GB of RAM, all connected with a high-speed, low-latency network.
Fundamentally, the nature of research has changed in the past two decades. Prior to that, most research still followed one of two fundamental research paradigms – Theory or Experiment. The demands of cutting-edge research have advanced to the point where Theory and Experiment, alone, can no longer carry us forward. Computation has, of necessity, become a co-equal pillar supporting the discovery process. Theory development requires tractability which is achieved by ignoring the multiplicity of interacting factors that are always present in the real world. Yet many of today’s challenges, such as Climate Change and Renewable Energy, are system-level problems wherein such simplification will not yield useful answers. Here Computation steps in to generate explicit, quantified hypotheses for test. For its part, Experiment has been traditionally grounded in statistical design and inference, themselves both forms of Theory that are strained or broken by modern realities. Once again, it is Computation that dissolves the barriers via expectation via numerically intensive methods. Lastly, some experiments, of course, are simply too dangerous to perform. In both cases, Computation enables us to do in cyberspace what we cannot or should not do in real space. At K-State, research is evolving steadily towards cross-disciplinary, large-scale problems dominated by teams of scientists from different departments (and frequently different institutions) attacking problems via linked computational models. This style of research, using large datasets to drive detailed models, which in turn yield insights for the next iteration of the process, is what Beocat was designed and built to do.
High performance computing (HPC) is an increasingly critical foundation for research across the disciplines, from improving the foundations of linguistic analysis to protecting lives through better bullet-resistant vests. At K-State, Beocat provides critical computational resources for millions of dollars in current research grants, and is a strategic asset on pending proposals to various funding agencies. Beocat, over the few years, has moved from being largely irrelevant to being a core resource for multiple research groups across campus. Currently, Beocat serves over 130 active users in a dozen departments across three colleges at K-State. Our current system is the largest research cluster in Kansas with over 1000 cores (see Status), and has been built with an official budget of $0, with users contributing different machines as resources become available. Staff support has consisted of one CIS undergraduate student (supported by a research grant from Civil Engineering) to administer the system, with support from CIS systems staff as needed. Computational research is critical across many disciplines, particularly for the type of cross-disciplinary “Grand Challenges” which our scientists are increasingly addressing. Beocat is a vital tool at K-State to help solve these challenges.
Beocat is used by a wide variety of groups, departments, and colleges at Kansas State University. Several of those groups are:
- College of Agriculture
- KDD Laboratory
- Department of Biochemistry
- Bioinformatics Center
- Department of Chemical Engineering
- Department of Computing and Information Sciences
- Mechanical and Nuclear Engineering
- Department of Physics
- SAnToS Laboratory
- Argus Cyber Security Group
Access to Beocat is free for researchers on the K-State campus. If you or your research group would be interested in using Beocat, please send inquiries to firstname.lastname@example.org.