The Flight of the Penguin?

Think of the penguin and you'll think of a funny bird on the ice floes of

Antarctica. While few of those black and white birds actually live there, the

picture isn't far off. They waddle clumsily, often in a long file, for tens of

miles through snowstorms. They're hardy, mature (50 mn years), and their

progress is terribly slow.

Tux the penguin, the colorful mascot of the Linux operating system, matches

his black-and-white counterparts in several ways. Linux is hardy and stable,

mature (15 years is the tech equivalent of 50 mn), and it's made rather slow

progress over these years.

I remember the excitement nearly ten years ago when we put together a Linux

"package" in our labs: download, assemble onto CD-R, tune it for the

486, burn a CDR. From that month in 1996, CyberMedia gave out Linux CDs every

year with one of our publications, and we also used the opportunity to switch

our company's communications and mail server to Linux. That was an old 486, a

former desktop. It later ran for years by itself-often without keyboard or

monitor, or even a UPS.

All this, and free? It seemed evident that Linux would occupy a key place in

the server operating systems world, pushing back other OSs, including

vendor-owned UNIX variants, and Windows. There'd be Linux across the

enterprise, the home, and of course the Internet.

A decade later, Linux probably accounts for between a tenth and a fifth of

server sales. Why not more?

First, systems vendor support came in late. And when it did, the initial

years saw it more as an offering to fill a gap and extend a portfolio. Tuning

and optimization came later. Even today, Sun's Scott McNealy says of partner

Red Hat: "We'll give on our servers if someone wants, but

we'll tell them that Solaris operating system has more performance

and features."

Sans history or commercial backing, Linux had no ecosystem around, including

trained engineers. All this kept CIOs wary enough to keep Linux away from most

enterprise apps, but interested enough to experiment with it-usually for

file/print, and mail and Web proxy.

Today, history's repeating on the desktop. It's being bundled with

so-called "Rs 10k PCs" to save the Windows license fees, but without

careful selection, tuning, optimization, often leaving a kludgy mess that makes

for an unusable desktop and a poor aftertaste-and an incorrect user conclusion

that there's something wrong with Linux.

Yet Linux is reportedly inspiring Microsoft, even if open source itself is

not. With Longhorn/Vista getting so complex to develop in the good old ways, the

Linux development model-of creating, testing and certifying a

"kernel" and then allowing libraries, tools, services and apps around

it-is a likely option for future development at Redmond.

I would venture to suggest that the Linux desktop is a lost cause, except

possibly for keeping up some pricing pressure on Microsoft. As a server OS,

though, it remains a great alternative. That's why CIOs are testing the Linux

waters today-as this issue's cover story says.

Prasanto K Roy