Unified computing is the key to the Tactical Edge

Why would running a unified computing platform at the tactical edge be a good thing?

First we need to explore what the tactical edge is — and why it’s important to the US military.

What is the tactical edge?

The tactical edge is the forward-deployed area of operations (AO) where the military deploys for combat, reconnaissance, or other specialized missions. 

It’s where our troops are:

• Engaging enemy forces, whether that’s with infantry, aerial combat, or drone warfare.
• Gathering intelligence using edge devices like sensors and drones.
• Establishing command and control (C2) by setting up combat operation centers to coordinate operations and make real-time strategic and tactical decisions.

How do our warfighters make their decisions?

In an ideal situation, our troops make informed decisions based on vetted intelligence and other collected data, situational awareness, their training and experience, and standard operating procedures (SOPs). 

Sometimes the data they need hasn’t been fully analyzed, and officers must fall back on risky gut decisions or wait for validation from rear command. 

What if there was a way to get our troops the data they need for their tactical decision-making, faster? Well, edge computing use cases make that a reality.

Edge computing technologies provide the compute resources our warfighters need to process information in real time from sources such as internet of things (IoT) devices on edge nodes.

Today’s powerful and transformative AI workloads are often destined for the tactical edge, too, multiplying the benefits of edge computing.

What’s inside edge computing?

The essence of any edge computing deployment is creating smaller and more efficient computing devices that still have the power to process data right where it’s collected.

Because they don’t rely on sending data back to a centralized server in a distant data center or public cloud computing environment, edge architectures don’t have to worry about latency and bandwidth usage. 

This is essential for military use cases, where equipment typically operates in disrupted, disconnected, intermittent and low-bandwidth (DDIL) environments.

Designing software for hardware limitations

Each edge device fulfills a specific need and gives the warfighter a different capability. 

Some are more secure than others, some process data faster, some fly, some go underwater, some connect to the network while some don’t, some you hold, and some you wear. 

But they almost all have limited processing power and storage capacity. Edge devices therefore need software designed specifically for the size and function of their hardware, that is lightweight, and optimized for performance. 

The multi-platform, multi-app challenge

Edge devices all run an OS, often Linux, with applications on top. But there’s no standard for the versions of software, operating system types, security controls in place, networking, and so on. 

A lot of military applications run only on Microsoft Windows platforms and cannot run on a Linux server, or are big applications that cannot run on a small form factor edge device. 

So in a real tactical edge environment you may see Windows desktops, Windows servers, and Linux servers supporting the various applications and legacy systems that are needed to complete the mission. 

The typical setup in a remote ops center at the edge is a laptop connected to a network with servers running as virtual machines (VMs) operating various applications. 

Source: army.mil

But with the limited resources at the edge, it can be hard to allocate enough CPU, memory, and storage to each VM — especially when the overhead of a hypervisor is draining resources and decreasing performance.

And that’s not the only obstacle. There are many different technologies you need to know just to manage the set of systems needed to operate, not counting any specialized systems that are mission-specific. 

With different operating platforms and edge devices comes an increased burden of operational complexity. That can be tough on troops who need to focus on the other mission tasks that they have to perform.

Scoping our requirements

So, how can we help our troops at the tactical edge? Clearly we need a single edge computing solution that:

• Supports the mission’s required OS and different kinds of legacy and modern software, without requiring units to bring more hardware or shoulder an increased management burden.
• Has a streamlined build and operating procedure, so warfighters don’t need to worry about software or configuration, and can instead focus on the data they need to make decisions.
• Works consistently across different hardware and operating environments, providing mission flexibility. 

The answer to these requirements? A unified computing platform, based on Kubernetes.

Introducing edge Kubernetes for unified computing

While Kubernetes was initially designed for cloud environments, it has a modular and extensible architecture that makes it ideal for running workloads at the tactical edge. 

Kubernetes is scalable and flexible, and doesn’t cause havoc on resources like VMs and typical operating systems do — lightweight K8s distributions are very happy on smaller form factor devices. 

The central advantage of Kubernetes is that it can serve as the universal bridge across all the military’s different mission applications across diverse operating systems and hardware platforms. 

It allows our forces to deploy software across different systems quickly and consistently. The applications will run the same way every time whether the application is installed on a small server in an ops center or on a Raspberry Pi. Our troops can therefore focus on mission tasks instead of fiddling with software. 

Importantly, Kubernetes can also support VM workloads, creating a truly unified computing environment. While it was originally designed to orchestrate container workloads, the open source KubeVirt project extends Kubernetes to schedule VM workloads too, even side by side with containers in the same cluster.

Cutting the complexity

The problem is that Kubernetes itself is complex. As one DevOps enginee told our researchers: “Our biggest issue is how complex it is to the layperson. That hasn’t stopped us; we are slowly but surely moving more and more stuff into production and Kubernetes. But it slowed us down dramatically.”

Managing Kubernetes clusters is challenging enough, but the complexity is even greater when you’re using Kubernetes to run a mix of applications (modern and legacy via KubeVirt), in DDIL environments, on servers and small form factor devices. 

This is where an enterprise Kubernetes management platform comes in to provide automation and standardization across Kubernetes operations.

What to look for in a unified management platform

With a platform like Spectro Cloud’s Palette, incorporating a VM orchestration tool like Virtual Machine Orchestrator (VMO), you can get that complete Unified Computing Platform that can manage both containerized (modern) and virtualized (legacy) applications and deploy those same applications to edge devices. 

Palette will help the military run Kubernetes securely and is tailored to their needs at the tactical edge, easing management from a single pane of glass. 

Notably, Palette Vertex (our government specific version of software) is the only end to end FIPS 140-2 certified Kubernetes management platform on the market today.

Palette consists of four main components:

1. Distributed Architecture which provides a decentralized management plane that enables local policy enforcement and the ability to scale to tens of thousands of clusters. We’ve tested upwards of 10k+ clusters and 30k+ edge devices under management without failure.
2. Full Stack Declarative Orchestration to model complete K8s stacks, from the OS to the application, providing a comprehensive blueprint (or template) that ensures compatibility between the software components running in the cluster.
3. End-to-End Lifecycle Operations providing a way to drive consistency, repeatability, and operational efficiency across multiple clusters and edge devices with comprehensive day 0 to day 2 management.
4. Flexible Deployment Model to deploy clusters to any environment including cloud, hybrid-cloud, on-premises, and many different types of small form factor edge devices — in both connected and airgapped environments..

A new era for the tactical edge is here

With a unified compute stack based on Spectro Cloud Palette and Kubernetes, the vision of tactical edge computing becomes a realistic proposition.

Teams at forward operating bases can benefit from a choice of compact and cost-effective small form factor hardware, and automate every stage of getting software onto that hardware to pursue the mission. 

As we’ve discussed, that software can include both containerized and VM workloads, orchestrated with the power of Kubernetes, in turn automated by a single management platform. 

To learn more about Kubernetes at the tactical edge, take a look at our webinar.