Linux's Success Pushing the Competition's Buttons

Six years ago, Jon Maddog Hall spoke to the DFW UNIX Users Group at a hotel in North Dallas. His sponsor was one Dave Whitinger, the founder of the then fledgling on-line news site, Linux Today. Jon gave two talks including one on how people once made clocks and the other on Linux.

During the Linux session, Jon discussed the D H Brown report on Linux. That was the big news of the day, because the DH Brown report gave Linus a blueprint on how Linux could become an enterprise operating system.

During the question and answer session, someone asked why not support the FreeBSD projects instead of Linux. Jon answered simply that Linux had more momentum. That sat well with many of the Linux special interest group members, but a few moans went out among other UNIX SIG members.

I had just started a pay-per-incident Linux call center and had an interest in the subject of the user base for Linux. Jon said that Linux had approximately 2 million users and the BSD's had 250 thousand. I had thought that the Linux population had to be larger than that.

Call Center's Don't Lie

One of the things that prompted my interest in starting a Linux call center resulted from my experience with Gateway.net, the ISP Gateway 2000 started for its customers. Gateway.net had problems with call failure rates and the company stakeholders brought me in to fix it.

Gateway had outsourced their ISP service to Web America Networks (WAN) in Dallas. The company used 100% Microsoft technology and it was not working well. In some areas we experienced call failure rates in the 80% range. WAN did not seem to know how to fix their problems so they continued to buy more Dell servers (that's correct) to expand capacity.

Mail had become a major problem and one of the engineers brought in a cobalt blue box, set it up and fixed mail. At the time, I did not realize that the blue box was a Cobalt Qube running a variation of Red Hat Linux 5.0 hacked to run on a MIPS processor. I did know, however, that mail came off of my list of problems.

In the meanwhile, the chairman of Gateway had started getting hate mail because of the ISP. According to my team, Ted had never gotten hate mail in his entire career. So, he was furious.

I finally met with the stakeholders in Sioux City to report my findings. I concluded that they needed to have WAN switch to Sun or go someplace else. The main problem Gateway faced dealt with getting the price of service down to the price point at which they sold their ISP subscriptions.

I suggested that we call UUNet because I had heard MSN and Earthlink provisioned their POPs from UUNet. That tuned out to be the case. UUNet came in and made a proposal that gave Gateway the room they need to switch in the event WAN wouldn't fix the problem.

UUNet suggested using a third party to handle operations. So, Gateway sent me to the contractor's office to perform due diligence. If they were going to switch, Gateway wanted an experienced person looking at the ISP's business processes.

What I found was a small company with a half-dozen Linux servers co-located in a little town in Pennsylvania. I wondered how such an operation could handle 2 million users and achieve call failure rates below 5%. I had just come from WANS where the data center covered the bottom floor of a large building and modem banks covered racks and racks of data center space.

To shorten the story, Gateway eventually went with UUNet and the small Linux shop and achieved their objectives. They did it with a small number of Linux servers. That convinced me to start bundling services for ISPs. So I started a call center and discovered Linux users abounding everywhere.

Figures Don't Lie, Liars Keep Figuring

That's a cliche I learned as a CPA when I got out of college and joined a local firm. We had clients that loved to use creative accounting to make their financials look good. But, in the end, everything boiled down to cash flow. If the company was putting cash in the bank, they were doing well.

Linux keeps putting servers on racks and people like Theo de Raadt can claim whatever he wants about Linux. Let's see, HP claims they've shipped 1 million Linux servers. The question remains how many of Theo's OpenBSD servers have they shipped?

I've used OpenBSD and sometimes have recommended it because at the time, they never had a documented break-in. But, that's not true today. Reading the interview with Theo de Raadt put a bad taste in my mouth for him and his interviewer who wrote the story. Hey guys, where's the beef?

People have started copying Linux technologies because Linux is the innovator and the leader in operating systems. From virtualization with Xen, VMware and Virtual Iron to Security Enhanced Linux and IBM's OpenPower initiative, I say Jon Maddog Hall made the right choice when he went with Linux.

Final thoughts

So let the figures speak for themselves. And the staff writers at ITNews.com.au and Computerworld.com.au who quoted Michael Sager saying "The Jury is in: Linux Has Not Taken Off on the Desktop"; well consider the CPA cliche. I would ask which Jury? Which court? Where? How? and Who? My figures disagree with yours, Mike. I didn't think this was a case for a jury.

Linux on the desktop has taken off in places in which you cannot do samples. In fact, no one can come up with figures for Linux desktops because no one can accurately measure the usage. If I download Ubuntu and pass the CD around, how will you measure that? How will you measure people dual booting to Linux? How can you measure people selling their copies of Windows on eBay and installing Linux instead? You can't.

IDC better come up with a realistic methodology to measure Linux desktop use or stop publishing reports about it. Of course, IDC could consult with Theo da Raadt. That's a credible source.