One problem with Linux is that nobody really knows how big it is. Like any other virus, you have no idea exactly how many bodies it has infected.

Ignore the sales numbers from Novell and Red Hat. They tell only part of the story, namely the demand by larger institutions who must ensure success and have support. These sales figures do not even come close gauging unsupported replications of subscribed distributions, hosted Linux (which often is self maintained), departmental servers, all the OpenSuse and Fedora installs, and the occasional renegade Linux laptop.

And those are the small markets.

I’ve been watching the embedded space more and more. Silicon Strategies Marketing has clients in the mobile phone business, the Linux business, and now in the embedded Linux space. We have been mapping where embedded Linux is finding traction, and some of the issues within that market.

The question is “where is embedded Linux not being used?”

• A relative of mine who works on Defense Department and “spook” contracts notes that Linux is the favorite platform for all military and intelligence embedded applications. As he phrased it “We can’t afford to reboot a spy satellite every few hours.”
• Linux is used in routers from a lot of different manufactures. Networking is a core Linux strength and with an embedded web server, it is easy to create user friendly interfaces.  And now everything needs to be network savvy.
• Phones are just starting to use Linux, but the open nature of the devices and the ability to plop new native applications on them is a strong differentiators. This is part of the reason Google went with Linux for G-phones.

Devices need to be smarter than in the past. This means they must have logic. It is far better to use an embedded operating system that has broad support. That really means Linux or Windows.

Linux wins mainly for two reasons:

1. It is more stable. Google wouldn’t run their entire product on Linux if it weren’t.
2. It is modular and tiny.

It is that last point that is perhaps most important. Much has been made of Window’s lack of modularity. WE (Windows Embedded) is considered by many to be a poor hack of XP, where fragility was induced by wholesale ripping apart of the operating system. I cannot comment directly, but this is the growing reputation. Some acquaintances of mine at Circuit City’s HQ said a WE deployment on cash registers was abandoned due to endless problems, which supposedly were induced by removing pieces of XP and seeing interwoven parts of the OS die.

Linux is by design module from top to tail. Much of Linux’s success has come by its ability to upgrade (or roll-back) discreet parts of the total package without disrupting other parts. This also means you can remove big chunks of unneeded functionality without much work or fear, creating custom versions of the OS.

This is why Linux will own the embedded space. When whittled down, the Linux kernel can be nearly 10oKB small, which is tiny. Total RAM requirements can be less than 4 MB. This makes putting Linux into nearly any device possible. Add the ability to customize all upper-level packages to operate well within small-device space, and you have something with which Microsoft cannot compete and simultaneously chase dreams of acquiring/integrating Yahoo.

Here’s the kicker: more smart devices are hitting the market all the time. Your dashboard (which soon will have integrated GPS navigation DVD player, Bluetooth interfaces, and more) is embedded. Your cell phone is embedded. Your home network is built on embedded devices. Your oven might require embedded logic to keep you from burning dinner … again. MP3 players need embedded OSs.

The list is endless and growing. The question then is “How do I profit from this?” That’s my secret for now.