InStat/MDR today offers some valid concerns concerning the rush of cell phone makers to Linux and Windows.

But they should have less concern over my advocating that makers of access points or mesh points build with embedded Linux or Windows.

Here's why.

The cell phone entries, which also include Symbian and Palm, are full-fledged operating systems. They provide a variety of services out-of-the-box that can, sometimes, interfere with normal cell phone operation, which is the whole point of a cell phone. (The picture of the inside of a corn seed, which shows you the size relationship between it and its kernal (not that the seed has a kernal but an OS has a kernel), is from Butte College in Oroville, California,

These operating systems also increase the price of the phone, which I think is deadly. A phone is not a computer. It is an interface. Turning it into a computer, with all a computer's problems, is a big mistake.

Using a robust, embedded OS kernel in an access point, or a mesh point like those made by Firetide, is another matter. A kernel is just a traffic cop. It defines how what will move where. It doesn't do anything, and it's not a service per se.

In my vision for these access points, additional services based on the kernel would be provided by additional chips. The chips will include all the software needed to provide each service. The kernel would continue to just play traffic cop. If there were problems they would be relatively easy to isolate. Unplug the application chip, and if the access point goes back to normal that is the problem. If you unplug the chip and the point is still giving trouble, then that's the problem.

It is vital, when thinking of Always-On, to think in terms of snap-on modules. That's a big key to reducing service calls and downtime, which in turn is an absolute must for consumer acceptance.

For a cell phone to provide the equivalent design service, it would have to be made of virtual Legos, and it would come apart. Maybe that's not a bad idea....