Being able to replace damaged modules while at sea or swap functions at the LRU level adds a whole new meaning to “SWaP-C.

When you’re tasked with maintaining technology on the battlefield or similar arduous environment, options are a good thing—usually. When your option is to pull a complete system and send it out for repair, well that’s likely not the best alternative, although that’s what regularly occurs. When your option is to swap in a new subsystem for a faulty one with a limited amount of reconfiguration, you’re likely going to take advantage of that option

The easiest way to implement the better alternative is by employing modular components. Modularity as it applies to mil-spec equipment allows you to remove a defective subsystem and simply replace it with a working unit in the field. This scenario sounds like a simple option.

Key Advantages Of Modularity

Configure Your System Exactly The Way You Need It
The first reason to go down the modular route is that it lets the users—in this case, the Army, Special Operations, the Air Force, and so on—configure the systems themselves with the right equipment. While they all might be looking to purchase a similar helicopter—take the Apache, for example—the compute needs of those groups will be vastly different. If you need more mass storage, you add this subsystem. If you need more I/O, you add that subsystem. If you need more video or graphics… well, you get the point.

Fixes In The Field That Even A Third-grader Can Do
With modular equipment, you can identify any problem without any special tools or diagnostic equipment: Simply unplug the faulty subsystem and plug in a known-good subsystem. The actual repair comes when you ship that faulty unit back to the factory. That’s a problem that no longer has to be dealt with on the battlefield.

Spares Made Easy
Assuming you don’t have a bunch of spare systems on board your carriers, you’re not likely to find a replacement after you’ve been deployed. If you’ve gone the modular route, you’re probably savvy enough to stock up on a few spare modules. But rather than have lots of spare modules sitting around collecting dust and waiting for a failure, you can pull modules from other active systems, which may not need a particular function at that time. In the heat of battle or while on deployment far from a depot, this scenario is attractive and common.

So Why Aren’t There More Modular Systems?

Modular Systems Cost More Up Front—Just Don’t Forget Total Cost Of Ownership
Sure, it costs less to build a system around a Supermicro server motherboard. However, if (when) that motherboard fails, the cost of replacement is quite high. There’s the cost of the motherboard itself, and there’s the challenge of gaining access to it among the rest of the system. It was never designed to be replaced. You’ve now far exceeded the cost of integrating the modular system. And at the end of the day, most people are headed toward a similar goal, which is driving down the total cost of ownership (TCO) on the battlefield.

It’s Hard To Meet Performance Requirements—But Not Impossible
Designing a modular system with the performance that’s needed for ruggedized/military applications is tough, but General Micro Systems has come up with a backplane design that can communicate between various system modules at a PCI Express (PCIe) Gen 3 rate. That’s subassembly to subassembly, not board to board. So clearly the design challenge isn’t impossible, for those willing to take it on.

How Did We Do It?

Ruggedized Flex Cable Is Key
To make this possible, GMS developed a ruggedized flex cable with integrated re-timers that amplify, clean up, and reduce jitter on high-frequency signals. It’s actually harder than it sounds when you factor in that all the subassemblies must be operating within pico-seconds of each other at what are literally RF speeds. In a fixed configuration, it’s hard, but not rocket science. But in a removable topology, let’s just say that such a design is not for the faint of heart.

S2U “King Cobra” Rugged Server Offers 100% Line Replacement Unit (LRU) Modular Subsystems
This technology is available in our S2U rugged server, which offers independent subassemblies that all link together with the high-speed PCIe Gen 3 connections. There’s a 6U OpenVPX backplane and chassis that houses the processing units and a switch/router function. That basically comprises the compute engine portion of the system. Next is a 3U OpenVPX chassis that can also be used for redundant power supplies or additional I/O.

A third modular assembly is a standard PCI card cage, with four slots that can accommodate full-size boards, with 16-lane PCIe each (making a total of 64 x1 PCIe lanes). The use of standard PCIe opens the door to deploy (or design) any type of card that’s needed—or the cage can be swapped for an auxiliary power unit (APU) with multi-minute hold-up. The platform’s modular drive array houses I/O modules or 12 SATA Express (aka PCI Express interface) NVMe drives, although you can use any type of drive.

The final subassembly is the cooling module, which is designed to operate in a rugged/military environment. Note that it’s all packed into a 2U half-height chassis, which is no small feat. It’s designed to go into applications where you historically couldn’t squeeze in this kind of functionality. For a similar feature set, you likely would have needed separate boxes.

The bottom line is, if you have infinite funds, feel free to stock lots of systems, and swap a new functioning system for a defective one. But if you are living under realistic budget constraints, modularity should be in your (near) future.

Ben Sharfi

CEO and Chief Architect
General Micro Systems, Inc.

This email address is being protected from spambots. You need JavaScript enabled to view it.