The point of cloud is to do more with less. To be agile and elastic. To put more money on mission. To break down data silos. To embed artificial intelligence and machine learning into our core warfighting competencies. To build new agile systems to keep pace with new threats. To get data out of the data center and push it to the edge. To make better, faster, and data-driven decisions. To use computing services to mitigate friction between military services. To bring data center capability to weapons systems. To bring this information and compute capability in a mobile, bring-your-own-device framework. And to knit sensors, supply-chains, satellites, and shooters into a single fabric to enable rapid and precise data exchanges across all domains to improve decision-making at the strategic, operational and tactical edges of war fighting.
With slightly different words, this aspiration has remained largely unchanged since the 1960s. But new technology paradigms offer opportunities to make real progress. Distributed systems helped us leap past the mainframe. Now, it is clear that proprietary, closed clouds were the necessary precursor of multicloud environments where hundreds of different services interoperate seamlessly in a single warfighting system. HAL 9000: “Sure, Dave, I can do that.”
Consumers today have an incredible number of streaming services (i.e., cloud services) to choose from, a far cry from the days when we relied solely on our cable provider’s predetermined bundle of channels. Today we readily move among multiple service providers, on whatever technology platform we prefer, and we bring our own device—a computer, tablet, smartphone, or smart TV.
Plug in an Apple TV and you can access Google’s YouTube, Disney+, Amazon Prime, Netflix, HBO Max, ESPN+, Hulu and countless other streaming services. Subscribe to all, some, or none of these services. Want to find a show? Search across all services. Don’t like Apple TV? Plug in a competitor and give Roku, Chromecast, or Fire TV a spin and access the same broad set of services. Competitors are cooperating and interconnecting, and customers benefit with choices and interoperability. We have become accustomed to expecting that everything just works with everything else.
Government expected a similar trajectory with the rise of cloud—more choice, more interoperability—but interestingly they have gotten less. Even though consumer and enterprise services are growing beyond “walled gardens,” federal departments and agencies have struggled to navigate the “walled forests” of cloud technology providers. As the government traded in proprietary data centers and on-premises technology for cloud they wound up with less interoperability, not more. To overcome interoperability issues, government tries to consolidate missions and systems into a single cloud vendor, irrespective of whether that is the best system for the task at hand. That can’t be the right answer.
The struggle for interoperability has become particularly acute at the Department of Defense and across the Intelligence Community where unique IT requirements are driven by operational needs. It is true that many of the services demanded by government users are the same services demanded by large commercial enterprises. Yet, a set of added requirements for redundant, dedicated, air-gapped, disconnected regions supported by Top Secret-cleared employees requires providers to heavily modify their commercial solutions for government customers. Unfortunately, the procurement and deployment strategies pursued by these agencies increase problems of single-vendor walled forests at the expense of interoperability and choice among best-of-breed technology services.
Cloud technology has expanded since its early days of basic—now commoditized—compute, networking, and storage. Taken together, the four U.S. hyperscale Cloud Infrastructure Providers (CIP)—Amazon, Microsoft, Oracle and Google—offer those standard commodities plus hundreds of highly differentiated and innovative services. Many hundred more cloud services are offered by companies such as IBM, VMWare, Cisco, as well as Slack, Splunk, Adobe, Atlassian, and Palantir (to name only a few). These capabilities run the gamut from artificial intelligence platforms, database technology, analytics, integration, collaboration and more. Because of this mix of commodity and highly differentiated capabilities, the best technical solution for a mission customer is to mix and match cloud services from multiple different vendors. This model of choice and interoperability at the services level is highly demanded commercially and is essential to meet the diverse mission needs of the Defense and Intelligence community.
Yet, for most in Government, multicloud merely means setting up a choice between more than one vendor and then operating within their walled forest once the choice has been made. The problem with this approach is that any given CIP will deliver some cutting edge, some mediocre, and some objectively bad cloud services. Some services excel at certain workloads (e.g., processing imagery) and some excel at others (e.g., high volume, complex mathematical computations). Why saddle a mission system to a less-performant offering when the workload, for example, requires a high-performance, low-latency service? Further, price competition can’t really occur because independent services are not really competing. The government’s unique mission requires a true hybrid, multicloud, multi-technology solution. The end-state of true multicloud is the ability for a customer to mix, match, interconnect, and interoperate among all these varied vendors’ services.
The imperative for cloud interoperability underlies the Department of Defense’s need for an agile and responsive future force that can work seamlessly across all warfighting domains, theaters, service branches, and with partners and allies. The future command, control and communication structure into which the Department is investing hundreds of millions of dollars will connect every warfighter to everything—from sensors to AI analytics to weapons platforms—everywhere, whether that be in a CONUS-based headquarters element or onboard a naval vessel in the Indo-Pacific theater. This Joint All-Domain Command and Control (JADC2) vision hinges on the cloud, as the Department notes when it speaks of its Joint Warfighting Cloud Capability (JWCC).
But JWCC and the Intelligence Community’s Commercial Cloud Enterprise (C2E) must deliver more than the option of choosing among hyperscale cloud providers with a set of accredited data centers in the continental U.S. and a tactical edge capability. Right now, key nodes in our defense and intelligence ecosystem (e.g., weapons systems) have been independently designed, built, and maintained without sufficient thought to how they could comprise a military “internet of things” running on an “internet of clouds.” The road from sensor to shooter starts with a vision for multicloud where systems participate in an interoperability fabric that knits mission systems together independent of the underlying technology.
A less appreciated reason to ensure service level competition is to capture commercial innovation that happens at the cloud architectural level, even while compute, storage, and networking are commoditized. Architectural innovation targets the entire stack of technology used to deliver cloud services and can result in better performance (lower costs), higher security, and a smaller required footprint for similar services. Any one of these qualities can be the deciding factor for delivering mission solutions to a forward-deployed warfighter in a region without built-out data centers. Cloud regions that can offer a full range of services in smaller spaces also require less power and cooling, which makes deployments on aircraft carriers, submarines, or on forward bases possible. There is a lot of white space between fully built-out CONUS regions and the tactical edge that requires attention.
It is clear the U.S. Government needs agile, resilient, and secure cloud services that interoperate with others. Some CIPs can offer more options for an agile, distributed cloud deployment, allowing the government to best match each deployment scenario to a vendor that can meet operational requirements and constraints. And by ensuring interoperability among the vendors, the Joint Force can remain fully integrated and agile while capitalizing on the benefits of an interoperable multicloud fabric with full access to the various services afforded therein—whether that be on a submarine or at a CONUS-based headquarters element.
To achieve interoperability, we need to address both technical standards and physical connectivity. And, as the DoD has done repeatedly and successfully over the decades, it now needs to use its purchasing power and take several steps to achieve the best answer for Government IT:
Mandate Interoperability: The government should mandate that vendors work together on common standards to enable interoperability across cloud services as a condition of participation in government cloud programs.
Open technical standards for interoperability are numerous and mature—the stack of protocols and standards for the internet form the essential connective tissue that enables modern consumer and enterprise services. Services should be interoperable from a variety of CIPs without the need to move data or engage in extensive (and expensive) migrations. Agencies should be able to try before they buy. Mix and match for analytics or AI. Or seek more performant solutions to speed workloads or save money (the other side of the metered cloud coin). A single mission should be able to easily use the cloud services that fit best from two, three, four or more vendors.
Mandate Interconnectivity: The government should mandate all hyperscale CIPs physically interconnect their clouds so that services can be accessed without moving or migrating data.
Physical connectivity among cloud infrastructure providers delivers network interoperability with significant benefits for customers and vendors alike. With high-bandwidth, low-latency network interconnects customers can leverage technical interoperability to compose mission solutions that seamlessly pass data from one vendor to another without having to downshift their processing tempo or incur extra costs and risks of copying data multiple times to run AI or other analytics. Numerous existing commercial agreements between CIPs for high-speed interconnects illustrate both the utility and market acceptance of enhanced physical connectivity.
New government requirements for secure high-speed interconnects among all federally qualified hyperscale cloud providers should ensure there are no physical restrictions for maximal technical interoperability. Because once the government selects which providers to work with, it is essential to tear down the physical connectivity barriers between them.
Mandate Open Data: The government should promote the free flow of data.
It is critical to end business practices that prevent choice and interoperability, such as data egress fees. In an open data environment, it is hard to imagine any vendor charging taxpayers to leave a particular service or to move data from a sandbox environment to an operational environment. As data is exponentially added to systems over time, egress fees grow and serve as a lock-in mechanism to prevent the movement of data out of one vendor’s system to another.
A clear federal policy that prohibits data egress fees is an essential part of ensuring both technical and physical interoperability requirements are leveraged to the maximum extent possible. It’s the only way to ensure the services required by departments and agencies are deployed, without worrying that funding for a mission system will be exhausted by simply using the data it creates.
It’s time to update cloud policy to ensure a successful multicloud ecosystem for the DoD, IC, and the broader defense and law enforcement community, and to our five-eyes and allied mission partners. It is clear the four U.S. hyperscale cloud vendors are all capable of providing services at scale and in secure environments. While these services must be certified and authorized to ensure security standards, it is also time to demand true interoperability and embrace open multicloud architectures across all capable vendors.
With heightened urgency, government policymakers must return their focus to ensure the largest range of innovative services are brought to the fight alongside the warfighter, intelligence officer, and law enforcement professional. By their own admissions, these communities recognize they must quickly adapt to the new operating environment—one of strategic competition and an imperative to maneuver within and control the information environment.
A modern, interoperable, multicloud capability is the technical backbone these entities must have to succeed. The U.S. Government has a long history of using its purchasing power to drive innovation and change. Cloud procurement should be no different—the government should mandate interoperability and interconnection to promote innovation and choice, all in service of the mission. Our national security requires us to do this right.