Primary tabs

Common Credits Model

[this is a cross-post from the NIH's Data Science blog - ]

By George Komatsoulis

One of the reasons that the BD2K program was created was to address three critical facts about biomedical research data: it is increasingly going digital, there is a lot of it and the amount is rapidly increasing, and it is difficult to find and reuse even the data that we have. The broad BD2K strategy is designed to address all of these facts as well as other important issues such as training, and communications, etc. A major effort within the ADDS office is The Commons, a scalable, shared, virtual space that exploits new computing models, is anticipated to be more cost-effective given digital growth, to simplify sharing digital research objects such as data, software, metadata and workflows, and to make digital objects more FAIR: Findable, Accessible, Interoperable and Reusable. Vivien Bonazzi has articulated a conceptual framework for The Commonselsewhere (see her excellent description and has discussed a number of pilot activities that are testing various elements of The Commons framework.

The Commons: Conceptual Model

At the base the Common’s framework described by Vivien is ‘Infrastructure,’ the compute, storage, networking and other core Information Technologies that are required to work with digital data and software of any kind. In the Commons, NIH plans to use (at least initially) public and publically accessible private clouds and high performance computing centers to provide these capabilities because:

Clouds are highly scalable, and most of the major vendors can easily support the expected growth of biomedical data over the next few years.

Data and software in clouds can be broadly accessible and shareable

Clouds offer the opportunity to have digital resources available in a single location [or available in ways that don’t require extensive local resources] and so can reduce duplication, which currently costs a significant amount of NIH dollars.

The economic model of clouds is ‘pay only for what you use’ which, combined with the reduction in duplication, increases cost effectiveness and democratizes access.

In addition, there are a wide range of public cloud providers that provide various levels of service(s), from simple infrastructure to development platforms to value-added software and do so with a wide range of price points. Ideally, this means that there is likely to be a provider that can cater to the needs and experience levels of every investigator, whether they are the power users that are building the capabilities that are the Commonsto computational novices that simply want to store data or use the tools that others have built.

In order for the Commons to be successful in the research community then, the NIH must be able to provide its funded investigators with access to a wide range of cloud providers, to ensure that those providers are accessible to all, and to encourage providers to make their services available at reasonable prices. In order to do so, NIH has concluded that it needs a new business model that can take advantage of (or even create) a competitive marketplace for biomedically useful IT services, as well as providing a convenient, cost-effective way for investigators to gain access to those services. Enter the Commons Credits Model.

The Commons Credits Model is actually quite straightforward. It envisions that the Commons Infrastructure component is a set of computational service providers, each of which meets NIH-defined requirements for access, capacity, interfaces, networking, authentication, authorization, and information assurance. Where the Commons Credit Model takes a new approach is in how it proposes to support this system. According to this new approach, funding for this infrastructure will not be distributed directly to these ‘conformant providers’, but rather to investigators in the form of ‘Commons Credits’, dollar-denominated vouchers that can be used to secure services from the provider of their choice.

The Commons Credit Model

According to the model, investigators will be incentivized to use their credits efficiently by selecting vendors that provide the best service at the lowest cost, while vendors will be incentivized to provide the best service at the lowest cost to attract a larger share of credits distributed to grantees. Leveraging BD2K and other digital object indices will provide the ability to find and ultimately use digital object within a cloud commons. while also reducing costs and enabling the NIH to obtain real data about how and which data and software tools are regularly used by the community.

The Commons Business Model Pilot is a three year, four-phase project designed to test this hypothesis:

Phase I - Creation of infrastructure: Finalize provider conformance requirements, arrange for initial providers and develop a portal to support investigators that want to submit requests for credits and select vendors.
Phase II – Validate infrastructure: Provide credits to a small number (20-40) intramural and extramural investigators with expertise in using cloud computing to test the infrastructure.
Phase III – Initial scale up: Provide credits to a larger group of NIH grantees, evaluate results.
Phase IV – Broader scale up: Broaden access to credits to NIH grantees.

Throughout this project, we intend to focus on being investigator-centric, ensuring that application processes and credit application procedures are fast and simple, and on bringing new capabilities to investigators in an agile fashion.

I am pleased to report that we have recently executed a contract with the Department of Health and Human Services Federally Funded Research and Development Center (FFRDC) managed by the Mitre Corporation to implement this pilot. The choice of the FFRDC was driven by the NIH’s need for a partner that can carry out quasi-governmental functions and has no conflicts of interest with regard to potential providers. The project was kicked off on October 14, 2015. The FFRDC will be working with us and the community of investigators and providers to finalize the provider conformance requirements, certify vendors, accept and review applications and distribute credits. Many of these processes are still to be defined (particularly the mechanism by which credits are distributed), and we want feedback from both investigators and vendors on the most efficient way to handle them.

If you have questions or comments please feel free to contact me at:

[this is a cross-post from the NIH's Data Science blog - ]