Amazon Web Services. Google. IBM. Microsoft. These are just a few of the major tech movers and shakers partnering with researchers and “big data” providers—recently, the National Oceanic and Atmospheric Administration (NOAA)—to invest in a new way of supporting a data-enabled economy: cloud computing.
The advantages and opportunities that come with working in the cloud are potentially significant for researchers, especially in terms of multidisciplinary collaboration, something CU Boulder’s Earth Lab team discovered firsthand after entering a cooperative research partnership with DigitalGlobe last September. The agreement allows Earth Lab researchers to access and work through DigitalGlobe’s 80-petabyte, cloud-based library of high-resolution satellite imagery, data and analytics tools.
The ease of access to powerful data on such a massive scale has proven a key catalyst as Earth Lab works to advance Earth and space science research alongside other pillars of CU Boulder’s Grand Challenge. The experience has sparked an inevitable question: How might cloud computing enhance and streamline the research being performed at CU Boulder campus-wide?
Terri Fiez, vice chancellor for Research & Innovation, has selected a team housed within the Grand Challenge initiative to execute a definition study exploring how research computing on the cloud might benefit CU Boulder and its partners in the future.
"Cloud computing has the potential to enhance existing collaborations and stimulate new ones between CU Boulder and its many research partners, both internal and external," says Fiez. "Discovering how the cloud can best support our researchers will be a key step forward in developing our long-term strategy as the innovation university."
While the need for high-performance computing (HPC) will likely remain in the coming years and beyond, a hybrid strategy that integrates cloud computing is quickly becoming a viable, and even vital, approach. Cloud computing delivers the same resources as a traditional data center at a lower-operational cost, allowing users to “rent” services on an as-needed basis without the upfront capital expense that comes from provisioning HPC resources.
The flexibility of the cloud platform also promises to maximize the speed, scale and collaborative output of research partnerships.
“Using virtualization approaches like containers in the cloud allows researchers to better collaborate with partners, since they are already using those approaches,” says Thomas Hauser, director of research computing for CU Boulder. “Containerized computational approaches enable CU researchers to create reproducible research workflows and share those approaches with our collaborators.”
Larry Levine, director of Information Technology for CU Boulder, says he expects the campus to eventually move toward a "cloud-first" philosophy—where the cloud is the default (but not the only) answer for investigators’ computing needs. The question is always: "What is the most optimized, efficient and cost-effective way to share data and manage access to that data?"
Levine says, "There’s no right or wrong answer. It will depend on [the] type of work people are trying to get done."