Best listening experience is on Chrome, Firefox or Safari. Subscribe to Federal Drive’s daily audio interviews on Apple Podcasts or PodcastOne.
The Army has taken a key initial step toward significantly overhauling the stovepiped and redundant mechanisms it uses to gather and disseminate data, moving toward a new framework it calls the Unified Data Reference Architecture (UDRA).
In a new request for information, the Army told vendors it’s looking for input on the new architecture, which aims to treat data as a product that soldiers with the proper authorization can access from anywhere, using cloud-enabled infrastructure connected through the concepts of “data fabrics” and a “data mesh.”
Officials have already built a digital model of what the Army thinks its data reference architecture should look like. In the Oct. 28 RFI, officials asked vendors to offer feedback on how they’d build and support something like it; responses are due by December 2.
Overall, the goal is to “flatten” and streamline the complicated data environment the Army operates in today. Young Bang, the principal deputy assistant secretary of the Army for acquisition, logistics and technology, said the service wants to move to a more federated and decentralized data environment.
“We really have to simplify and accelerate getting data to the necessary commander so they can enable their decision making,” he told attendees at the Association of the U.S. Army’s recent annual conference in Washington. “The fact that we’re insistent on replicating data at every echelon doesn’t work, and it clogs our network. And by the time a corps commander gets that data, it’s already multiple days too late — the enemy has already moved on. We’re doing things by exception to try to get the data to the commander faster, but it shouldn’t be by exception. So we’re trying to institutionalize these type of things.”
The Army plans to implement the Unified Data Reference Architecture (UDRA) in phases. To begin with, it’s focusing on analytical data products produced at relatively high echelons of the Army’s command structure: the corps level and above.
The Army wants to organize the architecture around the fairly new industry concepts that make up a “data mesh,” including the idea that “domain owners” should be responsible for particular types of data. In the Army’s case, those are defined, for now, as mission command, intelligence, fires and effects, maneuver, sustainment, protection and cyber defense.
“The key thing is we’re not trying to come up with a single data model,” said Dan Andrew, a systems architect in the office of the deputy assistant secretary of the Army for data, engineering and software. “Instead, we’re going to federate across multiple data domains to say, for example, ‘Warfighting system domain, you have responsibility for this data. You can govern it within that domain, but don’t be creating products and working on things that are in the business systems area. So we’ll be decentralizing, allowing the data to be owned, managed and stored within those domains, and not replicated and persisted into a bunch of centralized single points.”
Once those functional communities have governance and management authority over their data, the Army also expects them to take a product-centered view toward delivering that data to soldiers. Andrew said the Army wants to create a continuous feedback loop to continue improving those products to meet soldiers’ needs, but also enable a degree of self-service at the tactical edge.
“The data product has two key components: there’s the actual data payload, the content that the consumer needs, and then there is the metadata about the data. And that metadata has lots of information about it,” he said. “If a producer has product, they will publish that product out to an enterprise service as a catalog. But all they’re publishing there is the metadata about what their product is. It will be an API where consumers can go reach that data product, but that API endpoint is back in the the product producer’s domain, in their computing environment. So the consumers can go to the catalog, query it, look through it and say, ‘This is a data product that I believe meets my need.’”
Andrew said the initial RFI is focused mainly on the data products themselves. Future versions will help flesh out the specific definitions of the data domains, and the enterprise services the Army will need to tie them all together and make those APIs available to end users.
And there’s other infrastructure work that needs to be done too. The Army wants to move to an ecosystem of “transport-agnostic” data fabrics, making use of commercial satellite, cellular and any other network technologies it has at its disposal.
“We want robust transport, a cloud enabled-environment, and we want to use our data fabric as a catcher’s mitt for all the data being produced by all of our sensors,” said Maj. Gen. Jeth Rey, the director of the Army’s network cross functional team. “And the data mesh is going to allow us to get to where that data is and use those APIs across the board. And when we create data in the future, we’ll be tagging it, labeling it, so we have better understanding of it using AI and machine learning. And cybersecurity is going to be key. We’ve got to bake it in early, because we know that folks are going to try to attack our networks.”