Even as the Defense Department launches its much-anticipated Joint Enterprise Defense Infrastructure (JEDI) program, DoD Chief Information Officer Dana Deasy already is looking ahead to the ways in which the department might leverage the enterprise cloud platform to support key components of its broader modernization strategy.
DoD officially kicked off JEDI on Dec. 11, moving ahead with the program and its contractor, Microsoft, despite an on-going protest by Oracle. Between now and mid-February, the department plans to build out the unclassified environment, followed about six months later by the secret environment and eventually the top-secret environment.
But part of what makes JEDI important is that it will serve as a platform for DevOps, said Deasy, speaking at the Air Force IT Day, hosted by AFCEA Northern Virginia.
DevOps is a software development methodology that brings together the IT and operations teams to collaborate on defining requirements and testing functionality. The goal is to build, test and deploy software in small increments, both as a way to identify problems early in the development lifecycle and to get new capabilities into the field more quickly.
A number of software shops across the services have already adopted DevOps, and the department will look to them for ideas on how to roll it out across the department, Deasy said. One example is the Air Force’s Kessel Run Experimentation Lab, a high profile initiative that is using DevOps and other commercial best practices to accelerate the delivery of new capabilities.
JEDI also is seen as a critical platform for artificial intelligence-related initiatives, with the Joint AI Center (JAIC, pronounced “Jake”) poised to be one of the earliest adopters, Deasy said.
The JAIC, the department’s AI center of excellence, has proven to be more successful than DoD leaders expected, he said. Since receiving its first substantial funding in February 2019, the center has grown from a team of about 10 to a team of about 110. The JAIC has delivered several iterations of a predictive maintenance application to the Special Operations Command. It also has demonstrated the use of AI in predicting how a wildfire might spread.
“The JAIC is moving at a speed faster than we thought it would,” Deasy said.
However, the launch of JEDI does not mean that the department will be shutting down other cloud platforms. Having an enterprise platform will eliminate the need for many of those platforms, “but there will always be a need for special-purpose clouds,” he said.
“Now that we have started to stand up JEDI, it [allows us] us to look across DoD and challenge where we have siloed clouds: Do they serve unique purposes apart from JEDI, or is there overlap?” he said. “We won’t know until we actually stand up JEDI and bring it to life.”