The Agony and Ecstasy Of Becoming a Linux OEM

jammag writes "An article at the site Datamation, entitled Becoming a Linux OEM: A Roadmap, talks about the challenges (and rewards) of selling hardware with Linux pre-installed — most likely a growth market in the years ahead. The interesting part is the description of how some smaller Linux OEMs have made it. The bottom line: surviving as a Linux OEM requires far more than making it as a Windows OEM. In particular, you have to make the systems idiot-proof for users who don't care a whit about what OS they're using."

Comments

OEM=Original Equipment Manufacturer

[wehe]

The article titles one section "Linux OEM companies can survive, even flourish". In this section three US based companies are considered as Linux OEMs. At least for laptops and notebooks this statement seems wrong, because as far as I can see non of these companies manufacturers these devices themselves, though they pre-install Linux on them. Unfortunately there doesn't seem to be a really "free" laptop or notebook available yet. But at least you can get Linux pre-installed on laptops and notebooks from different vendors around the world.

Hell I can do that right now.

[khasim]

If Linux were more popular in the user community, how long do you think it would be before someone decided to email around a shell script that had some local privlege escalation code in it, and managed to work out a botnet from Linux boxes?

What's your email address? I'll send you one today.

The problem is NOT sending the email.

The problem is getting enough new people to run the script so that the infection rate exceed the disinfection rate. Otherwise the "virus" will "die" when it is "in the wild".

The more steps required to get a box infected, the lower the infection rate will be. Five steps at 90% success each ... still gives you only 59% of a successful infection. And that's not even counting the people who realize part way through that they shouldn't be doing that.

Sure the hole would get fixed relatively quickly, but the hacked computers probably wouldn't. We'd have the same problem that we have in Windows.

Nope. Microsoft's security model is badly broken in MANY places. Think back to their last anti-trust trial where they claimed that "integrating" IE's code with the OS was a "good thing".

All that means is that IE is a very easy avenue for OS cracks on Windows.

Try that with a Linux-based browser.

By default, Ubuntu ships with NO open ports. On Windows, they're all open, but they have a software firewall sitting on top of them. Again, Microsoft's security model is flawed. Which is why you STILL see cracked WinXPsp2 boxes.

Re:Is this FUD?

[James Wells]

Greetings,

There are looming possibilities of "patent claims" and "copyright infringement" against linux and the components that various distributions include. As a Linux OEM, I would think that fact poses significant risk to your business. It only takes one weird case/judge/lobby such as the old JPG copyright scares, etc to potentially put you at legal risk.

First thing to understand is that so far, there has not been a single proven case of patent infringement against Linux. Many people have claimed patent infringement against various packages on Linux and of those, there has been only two; MP3/4, which was IMO an unethical and barely legal patent, and DeCSS, though DeCSS wasn't really a patent claim when you get down to it.

Instead, what you have is someone like Micro$oft claiming that Linux violates their patents, but refuse to produce or defend the patents. You have people like $COX claiming that Linux violates their copyright, but refuse to demonstrate the violations, and when forced to by a judge, the judge effectively laughed them out of court. Please note that I am not saying that Linux doesn't violate any patents or copyrights, however, the simple fact is that, no one has been successful at proving that it does.

It should also be pointed out that there are quite a few companies who have come out and offered Linux both patent protection and patent amnesty, should it be determined that Linux is somehow violating their patents. This is the critical piece as most, if not all of these companies, are now donating code directly to Linux and the Open Source movement. Such notables as Novell, IBM, and SUN.

Finally, as a distributor, you have an ethical obligation to defend your clients from these patent / copyright claims, however, you also have the right to choose what packages you will distribute and support, but even more importantly you have the right to choose what not to distribute and support. One of the tricks with Ubuntu is that they tell you upfront that they do not distribute MP3/4 or DeCSS packages, nor will they defend their customers from claims in reference to these packages.

Re:THAT is the real last hurdle.

[Burz]

I agree that the community needs a good HCL.

But there's much more to it than that (see link above): You need a number of disciplines and structures in order to behave like a stable platform on PCs. If users don't see that consistency, and app developers aren't given a nurturing starting point (like Apple's XCode and ADC), and there is no clearcut way to distribute apps independently, then there will be a lack of top-notch applications to draw users to the OS.

Because we are not having this and many other discussions around LSB, because LSB isn't targeted by app devs, the software genre we fuzzily call 'Linux' just isn't a real computing platform. At least not one that is meaningful non-systems geeks, which is why the Linux genre tends to be only popular with sysadmins and system hackers. Users and the app devs that cater to them are still repelled.

Re:most people just want what works

[LWATCDR]

Actually I had a raid card from 3-Ware that didn't make into the Kernel from Suse so I had to compile it to get my server running after a kernel upgrade. It was a while ago but that is a good example of it happening. 3-Ware drivers not tend to be in the Kernel.
So yes here is an example of it happening. I had to call the hardware company and get the tar ball compile it and build a new kernel to get my server up.
It isn't just a problem with getting in to the kernel tree. The problem is that the hardware companies can not package the driver with the hardware! They can not control when the driver will make it into the distributions or if it will make it into the distributions. These are real problems that you dismiss in the name of GPL dogma. Notice I said that even if you FOSS your drivers you will have these issues. You don't like closed source drivers. That is fine because I also would rather have FOSS drivers. But has the lack of stable binary driver interface prevented closed source drivers? Gee I would say that is a big NO!