Defense CDO: Why market share matters in cloud

When it comes to all-purpose clouds, there are only three, according to the Pentagon's chief data officer.

"Generally they discourage us in government from mentioning the specific organizations, but I will be quite specific when I say the three big, general-purpose clouds are Amazon, Azure and Google," Michael Conlin said at a Jan. 16 event on artificial intelligence and automation hosted by ACT-IAC in Washington, D.C.

While Conlin wasn't making an endorsement, the distinction is useful for organizations to consider when making decisions about where to put their data.

"Everybody else is not a general-purpose cloud. They're a special-purpose cloud for unique things," Conlin said. "If you want to use them for those unique things, you have to be careful about the choices that you make. Because once you get in, it's really hard to get out. Just like putting something in your data center ... and I've seen data centers that look like archaeological digs."

Conlin told FCW that the companies with a large market share are more suitable for long-term, multipurpose use.

"Market share creates revenue, revenue creates R&D, R&D creates new services, and the pace of innovation in the big three is substantially above the pace of innovation in all of the others," he said. "We have to be able to move existing applications into this environment. The problem with all of the special-purpose ones is that you must convert to their preferred applications."

Conlin also said organizations looking to convert to a software-as-a-service model would fit with the special-purpose cloud offerings. "But if you have something that is not available in that environment -- and some of what we do [at the Defense Department] is not available on the software-as-a-service basis -- under those circumstances you need a more general purpose environment," he said.

Conlin stressed that his comments were similarly general-purpose, and not specific to any particular project or program.   DOD has hundreds of cloud initiatives, but currently is conducting two large-scale procurements that are attracting attention: The $10 billion Joint Enterprise Defense Infrastructure program aims to serve warfighters, and the $8 billion Defense Enterprise Office Solutions focuses on business applications.

DOD is in court defending its approach to JEDI. Oracle is suing the Pentagon, claiming the single-award solicitation process was unfairly tailored for a specific vendor. Amazon Web Services joined the lawsuit, filed with the Court of Federal Claims, as a defendant because of allegations directed against it in the complaint. A DOD spokesperson told FCW the JEDI solicitation process was still ongoing despite the suit, and an award is set to be made in April.

Using AI to help manage and organize DOD data was the backdrop for the CDO's comments and is also a big driver behind JEDI.

Conlin, who was previously Perspecta's chief technology officer, is charged with making sense of DOD's public and internal data. He said one of the organization's biggest challenges was cleaning up data.

Making DOD's data more easily accessible via application programming interfaces is also on his priority list. "We are a little behind on the API specs, but I'm trying to fix that," he said. The department has only published a handful of public APIs in accordance with Open Government Data policy.

Then there's aggregating, analyzing and managing information at the point of capture, which for military purposes can be difficult because "streaming data off of an aircraft makes a really good homing signal for incoming munitions," Conlin said.

"All the data that's being generated by sensors and scanners and all of these other devices is being generated out at the edge of the organization and actually needs to be managed and analyzed at the edge," he said.

Note: This article was updated on Feb. 4 to clarify that Conlin was not referring to a specific cloud program.