Linux is booming, but unpaid adoption may hurt vendors (Cnet)

Matt needs to explain the alarm he feels others should feel at having free deployments "nearly reaching parity in 2013", as if that was inherently a sordid state of affairs. And Matt needs to have more multi-sentence paragraphs.

Of course nonpaid Linux deployments are out pacing paid Linux deployments. Is anyone really shocked with this news? Sometimes Matt Asay hits the nail right on the head, other times, I think he tries too hard.

While Linux adoption into the datacenter has gutted Solaris, AIX, HPUX and the other "heavyweights" in the Unix arena, the adoption is still young, and the field is vastly open for anyone to make innovative moves, and make tons financially.

The issue at hand isn't so much purchasing licenses to USE the software as it is purchasing contracts to SUPPORT the software. This is why Canonical will be eventually successful, even if they haven't turned a profit yet (a problem of outgoing expenses and overhead, not necessarily incoming sales- they are brining in $30 million annually).

RHEL and SLES/SLED, while popular now, and will continue to grow in popularity, will be the minority in Linux deployments. I see a couple reasons for this:

* First, many companies just have their own support staff. The only time you need to call a vendor for support, is because you can't fix the problem yourself, whether you lack the staff or the know-how. If you have both the staff and the experience, support contracts are just fuzzy-wuzzy-feel-goods for suits, so they can sleep at night, and don't mean much to those actually working on the system responding to the pager at 2 in the morning.

* Second, if you work at a massive organization like I do, you eventually realize that you spend millions on licensing for all sorts of software. When you ask the tough questions though, especially in a trying economy that we're currently facing, you begin asking why, and start looking at alternatives.

For example, we pay tons for NetBackup licensing for our backup solution. We have 4 tape LT04 tape drives, and send tapes off-site regularly. Why NetBackup? Why not Project Amanda? When we started comparing feature set for feature set, we realized would could fully utilize Amanda, and not skip a beat in our daily routine, thus, saving us a good chunk annually.

Sure, there will be companies that cheat the system. There are always the dishonest, but they're also always the minority. If I were Red Hat, I would start looking at how to leverage free with paid revenue, possibly following the model of Canonical. Chasing after those who are cheating the system is the same mistake Microsoft and others make.

Lastly, Matt mentions that Red Hat spends $100 million annually in developing the Linux kernel, Fedora and RHEL. He further seems to conclude that if unpaid Linux distributions outpace paid distros, that it will be harder to develop the kernel and other software pieces. I don't buy this argument. Call me naive, but paying developers to develop the Linux kernel will always exist as long as companies exist. IBM, HP, Oracle, Red Hat, Novell and many, many others have paid employees or contractors specifically hired for developing the kernel. If the world were wiped of corporations the world over, or at least of all the tech-based ones, then maybe there should be cause for alarm. However, at the current moment, and probably well past 2013, I'm not worried.

Sorry Matt, this isn't one of your better pieces.

This article makes the same fundamental mistake that people often make when talking about the music industry. Which decreases the music industry's net income more, people who haven't paid listening to music industry music or people who haven't paid not listening to anything at all? For a long time (particularly since they started heavily promoting new artists instead of maintaining established artists), it's been the case that the main reason people buy music is because they've heard it already in some context where they didn't pay for it.

The same principle applied to software deployments; unpaid deployments don't cost the vendor anything, and are more likely to start paying than sites that use a different product or nothing.

In the limit, of course, RHEL would be perfect, and nobody would have any reason to pay for support (either getting bugs fixed or getting requested features implemented); on the other hand, Red Hat would also have no need for developers, since the product would already be perfect, and everybody would go off and work on designing flying cars or something. As long as we approach this state from the direction where people who care about Red Hat's progress are willing to pay more than what it would have to cost to them but less than what it's worth to them, the unpaid deployments are more of a benefit to Red Hat than those deployments not existing would be.

CentOS already overtook RHEL 18 months ago. There are now at least 4.5 million unique CentOS systems updating from CentOS mirrors (in reality probably threefold that amount).

Since CentOS can be downloaded and copied it is harder to find real numbers compared to a paid vendor with support. So we have to start from bottom-line numbers based on security updates and a guesstimated multiplier. I'd say at least 13 million CentOS systems worldwide., probably more.

I don't believe that for a second. Non-paid deployments must be at least an order of magnitude larger than paid. There are about a bazillion web servers out there for large hosting companies and they hardly pay for that. And there's no change massive companies such as Google and Yahoo pay for theirs. Then there are all the home and non commercial users. After all, Linux distributions all started in as non-paid.

This is good news for Red Hat.

I support a product typically deployed on Linux, the most commonn flavor is CentOS. If the (mission critical) system is encountering kernel panics, they call me thinking it's the application. I direct them to their operating system vendor.

At this point, they start wondering if they should have an operating system vendor.

This situation is not at all uncommon.

Sure if you're mr smartypants and can go surf around on the kernel mailing list etc then you don't need the vendor. A *lot* of IT workers aren't. They'll roll out CentOS because it's free, but it's likely they'll find they need support sooner or later.