Computing Technology for Research
Information technology plays a key role in almost every field of research. Whether crunching large computational problems and simulations, storing and processing large amounts of data, or exploring machine learning and artificial intelligence, modern research and discovery relies on computers more than ever before. Let's summarize the services available to researchers and how they can be used.
Compute Clusters
If your problems are too big to solve on your average desktop or laptop computer, you may need access to high performance servers work that have dozens of CPUs, large amounts of memory, or advanced graphical processing units. We manage two compute clusters called Nova and Pronto that include hundreds of high performance servers. Researchers purchase hardware that gets added to the shared pool of computers. Users submit jobs to the cluster job scheduler which assigns the jobs to computers that can satisfy the job requirements. Because users can run multiple jobs simultaneously, they can utilize much more processing power than they purchased, and get more work done in less time.
Data Storage
One of the biggest challenges in any type of research is data storage. Researchers consume a lot of data and produce a lot of data. A large capacity storage system is critical.
For longer term storage where capacity, cost, and reliability are more important than access speed, we provide the Large Storage System or LSS. Researchers and other organizations can purchase storage on the LSS for a relatively low price. Users may request from a few terabytes to hundreds of terabytes of storage capacity. Data can be accessed by the researchers as a regular Windows file share using standard file sharing tools. Data can also be transferred across the Internet using tools such as Globus.
High speed storage is provided on HPC clusters as well. Cluster storage is served over high speed networking to the compute nodes so jobs can run fast. While not as easy to access as LSS files, users can easily upload and download data to HPC storage as needed.
Applications and Development Tools
In any research, software plays an enormous role. We provide hundreds of software applications that are used in the physical sciences, life sciences, statistics and mathematics, fluid and structural mechanics, and machine learning and AI. We use the Spack package management system to expertly build and manage hundreds of open source tools and applications. We also offer commercial packages such as Matlab and Ansys. Environment modules are used to make it easy for users to access the specific software they need.
Inevitably, researchers have to write software. A variety of development environments are available for many languages including C/C++ and Python. Libraries for AI, parallel programming, and hundreds of open source libraries are available as well.
One of the most useful capabilities is the OnDemand system which allows users to launch a graphical interface running on a cluster compute server. This makes it possible for users to interact with applications graphically in addition to submiting jobs from the command line.