Howard Powell, Technical Director, Research and High Performance Computing
The Research Computing (RC) team has had an exciting summer working to grow our services to the research community here at Colgate. There are several projects to highlight, including an expansion of our virtual server and storage equipment, and a major investment in high performance computing equipment designed to process large quantities of data in a short period of time.
The new Virtual Machine Infrastructure enables us to “spin up” individualized servers with the click of a button. Replacing the long- standing model of individual physical servers dedicated to individuals or groups, the new system allows us to create servers without needing to dedicate new hardware (and without incurring additional costs against research funds and grants). These servers can be used temporarily or long term, and feature large amounts of memory and processor power — or “cores” — to meet the needs of nearly any project.
These virtual servers can be accessed remotely, making them easy to manage and configure by either the researcher or the RC team. Teams working together on a project have full access to the server and software they need, and student and faculty collaborators can remotely log in from computer labs or personal computers anywhere in the world. This virtualized system reduces the costs of research since we don’t have to buy as many physical servers or lab machines, we don’t have to power them and keep them cool, and we don’t have to waste precious lab space on computers; all while still giving you the computing resources you need to get your work done.
Supporting the new virtual servers is a fast, new, enterprise-class storage system. Featuring a 160 terabyte (TB) capacity, this provides the primary storage for faculty virtual servers, and can also provide networked storage for stand- alone servers and desktops.
For faculty with large data sets of up to 25 TB or more, we’ve identified cloud providers who offer inexpensive and reliable storage options that work with Colgate-provided servers. A wide variety of solutions exist for researchers who need a few gigabytes of easy-to-access space, up to about 25 TB of storage space for big data users, at costs ranging from free to less than $2,500 (one time cost). Colgate’s main high performance computing cluster, which is used for quickly performing mathematical calculations on massive data sets, hit 5 years old this summer. Colgate invested $160,000 in a new cluster, named for 20th- century computer scientist Alan Turing, to meet the needs of researchers with complex high performance computing needs. For those who like technical specs, the new Turing cluster has 20 nodes, 640 CPU cores, two TB of RAM, and about 13 TB of local storage. The networking within the new cluster is 50-100x faster than the network to your desktop. The Matlab software
is fully licensed for the entire cluster, as well as many other typical programming languages such as C, Fortran and Python. The measured performance of this new cluster is just shy of 19 TFlops. Benchmarking and tuning is happening right now, with faculty testing beginning this fall.
You can read more about the new cluster here: https://research-computing.colgate.edu/ tiki-index.php?page=Turing+Cluster.
On the advice of our Research Computing Advisory Committee, we’ve created a wiki-based portal with details on resources for faculty and students. Check out the work in progress at research-computing.colgate.edu.
Finally, over the next few months, Research Computing will be moving to a new space — Case 350. This area is near the main door to the Case Library and Geyer Center for Information Technology, near the elevator. We hope that this environment will be welcoming to faculty and student researchers and that you’ll come visit us once we’re settled in.
Feel free to reach out to us anytime at