Data is the engine that powers agency IT modernization efforts, powers improvements in customer experience and, of course, powers better cybersecurity.
The challenge for all agencies is the ability to harness that data and how they can use the cloud to drive better and faster decisions. Agencies need help to break down information silos and ensure secure data sharing is the rule, not the exception.
One example of this comes from the Veterans Affairs Department. It is creating a new Master Data Management system. The goal of that back-end system is to take data elements, create a centralized data repository for the entire agency and then share them across the department.
Determine your governance approach, whether centralized, decentralized or federated
As agencies continue to develop their data-centric approaches, there are several considerations to use the next generation of technologies to improve data management, said Winston Chang, federal chief technology officer at Snowflake.
“All of the infrastructure that is in the cloud has now been used and some of the first order problems have been attacked and understood,” Chang said on the Innovation in Government show. That work led to abstractions built on top of the cloud, which in turn has led to the development of purpose-built services ideal for difference federal uses, he added.
The end result? Data management now is seamless across different cloud vendors. “It essentially can bring a network effect to how data is utilized,” whether within a single organization, across an agency or event at locations globally, Chang said.
He shared two approaches agencies can use to take advantage of cloud to tap data to inform their decision-making.
Tactic 1: Tie data and applications to mission functions
Agencies should focus on the data layer instead of the application layer of the network architecture, which will simplify how an organization ties data and applications to a specific business function, he said.
“A great example of something that’s been very hot recently is the low-code, no-code development. It’s because it allows an abstraction of not having to do full stack development, where you really can just align to your business need,” Chang said. “Similarly, when we deal with data, which is sort of our bailiwick, the abstraction allows the architecture, the data management, the governance and all of those things to be closer to how the agency needs to operate and less about the technical things.”
The other benefit of abstracting the data layer is better experience across multiple cloud instances. Agencies can then focus on solving problems and not worry about where specific applications and data sets reside, he said.
Expanded government use of software as a service (SaaS) is also leading to abstraction of data. “So much can be automated, and it can be automated at scale — all those processes, all the infrastructure — agencies can just pay for as a service,” Chang said. “You don’t have to build it,” which makes that initial piece of the build versus buy discussion simpler too because there’s zero infrastructure cost.
Tactic 2: Enhance collaboration, vertically and horizontally
With the move to SaaS and by abstracting data from siloes and legacy systems, agencies can improve internal data sharing initially and then externally with other agencies and partners, including suppliers. They can also bring in open source data to further the understanding of agency mission needs and to help inform decision-making.
It can also aid in collaboration so that agencies make decisions at multiple levels, Chang said. By layering in code, an agency can tailor who sees what and how different organizations and people within an agency interact with the data and one another.
“This can work for the Defense Department. This can work for Department of Commerce or for any agency within the federal government. They can have those fine-grained, fine-tuned definitions, which allow for the collaboration to be optimized,” he said. “These sharing policies and the sharing can really reduce the risk of sharing because you can put on top of it policies that automatically check permissions — and you can watch the metadata that moves back and forth.”
This approach to collaboration can speed decision-making because “we can actually reduce the process procedure policy piece,” Chang said. “That completely changes the entire game.”