Windows or Linux?

r

Windows or Linux?
r

r

By Doug Kaye
r

r

This story appeared in the July/August 2004 issue of Web Host Industry Review magazine. Click here to subscribe for free.
r

r

July 29, 2004 — It’s the first question
r

you ask: Windows or Linux? Some hosts specialize in one platform, while
r

others offer both and attempt to present them with equal opportunity.
r

r

The Apache-to-IIS ratio has risen to 4:1
r

according to Netcraft, but the trend appears to have stalled since the
r

beginning of this year. By the end of 2004, however, I believe the gap
r

between Windows and Linux will resume its widening, and a few years
r

down the road I expect Linux will become the de facto standard hosting
r

platform. I see two causes of this trend: first, the economics of the
r

LAMP platform (the all-open-source combination of Linux, Apache, MySQL
r

and either PHP, Perl or Python), and secondly, though it won the
r

browser wars, Microsoft no longer wants you to develop Web-based
r

applications.
r

r

For the past year and a half, Doc Searls,
r

senior editor of Linux Journal, has been preaching what he calls
r

Do-It-Yourself IT. DIY-IT isn’t the same as open-source, the latter
r

being more involved with licensing and community. Rather DIY-IT is a
r

new take on the old buy-versus-build strategy. Doc believes that as
r

software becomes more of a commodity, like building materials in the
r

construction industry, power is shifting from the software vendors to
r

their customers, and it is becoming increasingly viable for customers
r

to build their own solutions rather than buy them from vendors.
r

r

In the late 1990s, the supply and demand
r

ratio for programmers was quite abnormal. In some regions, most notably
r

Silicon Valley, it was virtually impossible to hire top talent at any
r

price, and retaining employees was equally as challenging. An IT
r

manager had to make do with a smaller staff with a lower skill set, yet
r

still had to pay top dollar. The result was that vendors could make the
r

case for packaged software products that they’d never be able to sell
r

under normal circumstances. Open-source software wasn’t economically as
r

attractive as pre-packaged solutions because open-source projects were
r

more labor intensive, and labor was in short supply. Doc Searls might
r

refer to this as the pre-fab era of software construction.
r

r

But in 2000 the pendulum swung the other
r

way. Almost overnight, it became much easier to find great people who
r

had learned a tremendous amount about Web-based development during the
r

dot-com boom. Where there had been a dearth of programmers with even
r

basic skills, now there were legions well trained in the open-source
r

arts. Once again DIY became practical. At the same time, the LAMP
r

platform began to stabilize and standardize, making the build option
r

look even better in comparison to still-expensive proprietary solutions
r

such as the Microsoft platform.
r

When an open-source programmer moves from
r

one job to another he can often take his personal set of tools and
r

scripts with him, and when you hire a Linux programmer, he often
r

arrives with a complete toolkit. In the Microsoft world, the employer
r

usually supplies the tools because they’re not free. In a world where
r

tools and components cost real money, most choices are exclusionary:
r

you select A or B. But when they’re free, you can have them all. You
r

get to use “the right tool for the right job” instead of making do with
r

a bad choice just because you’ve already paid for it. The economic
r

advantage of the LAMP platform continues to grow two dimensionally
r

because all of the tools are freely available to an increasingly large
r

population of developers.
r

In addition to the natural gravitational
r

pull toward a free and standardized platform, I expect Microsoft to
r

hasten its own exit from the Web hosting marketplace. In an essay
r

entitled How Microsoft Lost the API War, Joel Spolsky points out that
r

Microsoft is discouraging the Web as a platform. In order to survive,
r

the company needs to encourage rich-client architectures (like Windows)
r

in lieu of Web-based applications. As Joel points out, Microsoft has
r

essentially abandoned enhancements to DHTML and other browser-based
r

technologies. He writes, “The big meme at Microsoft these days is
r

‘Microsoft is betting the company on the rich client.’ You see that
r

somewhere in every slide presentation about Longhorn.” Microsoft is
r

wooing developers away from the Web and back to the desktop because
r

it’s the only way the company can make money, and why I believe it will
r

ultimately lose the hosting-platform business.
r

r

Five years from now, will we continue to
r

see parity between the Microsoft and open-source platforms? From
r

reading and speaking with some very smart people like Doc and Joel,
r

I’ve come to the conclusion that the trends are not stable, and that we
r

are, in fact, approaching a tipping point that will ultimately shift
r

the hosting market substantially toward the open-source platform.
r

r

So why has the Apache-to-IIS ratio been
r

flat for six months? Perhaps it’s due to the litigiousness of SCO and
r

the FUD it has created in the marketplace. It’s remarkable how much one
r

small vendor can damage an entire marketplace. But make no mistake
r

about it: the SCO situation is temporary. I hope it will be resolved by
r

the end of this year, but it could drag into next. In any case, once
r

SCO goes away, the curves will again diverge.
r