NIH creates cloud services marketplace for researchers
- By Stephanie Kanowitz
- Nov 21, 2016
A pilot program is letting biomedical investigators get better access to data and computing via cloud services vendors vetted by the National Institutes of Health.
Through the Commons Credits Model, NIH approves providers that meet its requirements and standards, and investigators can apply to receive credits they can use to purchase cloud services from their choice from among the approved vendors.
About a year into the three-year test, three vendors have been certified, and an initial batch of investigators have received credits to work with those vendors. The model is part of a larger NIH project called Commons, which has the goal of making digital needs findable, accessible, interoperable and reusable, said George Komatsoulis, senior adviser for data science technologies and the Commons Credits Model at NIH’s Data Science division.
“We want to provide the maximum return on the taxpayers’ investment in biomedical research, and one obvious way of doing that is to ensure that data gets used for more than the usual purpose for which it was collected,” he said. “You therefore wind up with more scientific and health knowledge.”
The model is different from the Federal Risk and Authorization Management Program, which provides a standardized approach to security assessment, authorization and continuous monitoring for cloud products and services that agencies seek to use.
“FedRAMP is all about making it easier for government agencies to use cloud computing by allowing you to accept another agency’s certification and accreditation package, whereas in this case, we’re interested in getting things into public clouds and having these cloud providers compete for the business of potentially thousands of NIH investigators,” Komatsoulis said. We’re working to “incentivize them to provide biomedically useful services at the lowest possible cost.”
Biomedical research, like many areas, is becoming increasingly digital, and doing so at a pace so rapid that compute, storage and networking technologies are having trouble keeping up with the amount of data created, he said. This increase in data is also making moving, storing and sharing data costly at a time when funding is stagnating.
The Commons Credit Model attempts to address that problem of increased need and diminishing resources. NIH is implementing the pilot via a contract with the Centers for Medicaid and Medicare Services’ Alliance to Modernize Healthcare (CAMH) Federally Funded Research and Development Center (FFRDC).
Vendors that want to participate must offer infrastructure as a service, platform as a service or software as a service, and they must make the services available to the broad community, which right now means NIH grant holders. Additionally, they must meet requirements for interfaces, data deposits, download management, networking and connectivity, information assurance, compute and storage capacity, as well as the ability to launch computational tools for the data.
Desirable, but not required, features are related to continuity of operations, alternate interfaces, data encryption and free or reduced data download fees. NIH “is not providing any funding or resources for groups to become conformant. This is a business decision they have to make on their own,” Komatsoulis said.
After a vendor files an application with the CAMH FFRDC showing that it has met the requirements and has been approved, company officials must sign a participation agreement to receive the credits.
The process for would-be customers is similar. They apply to be part of the program by explaining what they propose to do and how it will benefit the Commons specifically and science generally. Subject-matter experts at CAMH FFRDC review and score the application and recommend to NIH which applicants should get credits. NIH makes the final decision. Selectees then review the vendors via the model’s portal and tell NIH which service provider they plan to use and how many credits they will need. The center arranges for the credits to be attached to the vendors’ accounts.
NIH provides no additional credits once an investigator uses them up.
It takes about two and a half to three months for investigators to get approved and about a month from the approval date to start working. NIH is keeping the credit distribution to no more than $10,000 for now.
NIH receives reports on the kinds of services investigators use the credits for – knowledge that will benefit NIH, Komatsoulis said. “Another valuable thing that we’re getting from this is an understanding of what it is scientific investigators need when they do research. Is it mostly storage? Is it mostly compute? Is it specialized services?” he said. “Right now I don’t think we have a very good handle on that.”
Only a third of the way through the three-year pilot, Komatsoulis said the remaining time will be devoted to bringing in more investigators and vendors.
“Now that we think we have a system in its initial operating capability, over the next 10 months it’s our plan to slowly but surely open the credit application process to larger and larger cohorts of investigators,” he said. The next round is slated to open in December, and ultimately the model would be available to anyone with an NIH grant.
Along the way, the agency will study metrics on what kinds of datasets and tools have become more readily accessible, whether investigators think the model works well and whether it is cost effective. If it is, NIH will look to more tightly integrate it with other grant programs, Komatsoulis said. “If not, well, then we’ve learned something useful in any case.”
Stephanie Kanowitz is a freelance writer based in northern Virginia.