Thanks to new computing infrastructure acquired by the Center for Data Science (CDS), Professor David Jensen and his research team are proceeding with a project to improve the supply chain of a major aerospace manufacturer. Aerospace manufacturing is extremely expensive, time-consuming, and complex. Wouldn’t it be great if there were a way to predict what manufacturing processes are needed and when? Helping to optimize the supply chain in this way is the work of Jensen and his team of computer science graduate students, in a project sponsored by a CDS partner company.
As their research began to mature, Jensen and his team ran into two roadblocks. First, the partner company was not easily able to furnish the high-performance computing infrastucture needed to perform the research and experiments that the project required. Second, the company was hesitant to transfer sensitive business data to the researchers without some assurance that the data could be protected at a higher level than is typical for most university research computing systems.
Enter the Center for Data Science. CDS leaders knew there was a solution to be found amongst the partner company’s business issue and restrictions, and UMass’s researchers and resources. Using grant funding from the Massachusetts Technology Collaborative, and with agreement from the partner company, CDS built the Secure Computing Enclave (SCE) -- a dedicated compute cluster, housed at the Massachusetts Green High Performance Computing Center (MGHPCC) in Holyoke.
At the MGHPCC facility, users must pass through multiple access-control levels to get into the building, into the data center, and into the cabinets where the SCE resides. In addition to physical access restrictions, the SCE has a one-way door of sorts, so that data can be easily transferred into the environment, but data can not be freely transferred out. Jensen and his research team can access the SCE remotely, and make use of the suite of CPU and GPU compute nodes, to perform research and run experiments. The SCE is set up so that data transferred out can only go to one place -- a secure server accessible only by personnel of the partner company. So, the results generated by Jensen and team can be moved off the SCE to a secure server, where it can be screened by company personnel to make sure no sensitive data is included, then made available for the researchers to collect.
When Jensen’s research is completed, the results could mean hundreds of millions of dollars in savings to the partner company. Now that the SCE has begun showing its value for this project, CDS plans to make the SCE available to other industry partners who have similar concerns with sensitive data.