The little OS that could

Lately, 10-year old Linux has garnered a lot of attention. There are some very good reasons--millions, in fact--for this. Linux kernel maintainer Alexey Kuznetsov gives a few pointers on how Linux can survive another ten.
Written by ZDNet Staff, Contributor
Lately, 10-year old Linux has garnered a lot of attention. There are some very good reasons--millions, in fact--for this. They can all be summed up in a few words. Linux is the right technology at the right price.

This is true for large companies and for small- and medium-sized enterprises (SMEs). The challenge lies in getting that message out and providing the tools and support SMEs need.

Fortunately, the message is becoming louder. Amazon.com recently announced that it saved US$17 million with its switch from proprietary Unix to open source Linux. The 25 percent cost reduction arose in part because of "the migration to a Linux-based technology platform that utilizes a less-costly technology infrastructure."

A year ago, IBM said it would plow a billion dollars into Linux. The company recently donated US$40 million worth of software tools to the public domain as a partial fulfillment of that vow.

There's also been many good reasons, thousands of them, favoring Linux in the small and medium enterprise arena.

A study by independent consultant Rob Valliere in September detailed how a business with a network of 25 computers upgraded from Windows NT to Red Hat Linux and Star Office instead of Windows 2000 and Office 2000. Valliere reported that the Linux-based approach was much less demanding in terms of hardware, and that the "primary reason for this decision was a US$10,000 cost saving."

It has grown up

Behind these stories lie some fundamental technical advances in Linux itself. The little operating system that started out as a hobby aimed solely at Intel 386 processors has grown up.

With the January, 2001 release of the 2.4 kernel, Linux added support for many new peripherals and processors. The OS also gained the ability to handle larger amounts of memory. Both of these are important in enterprise computing.

Plans for future kernel releases could include such key enterprise computing elements as symmetric multiprocessing scalability and file journaling.

But even without those enhancements in place, the current Linux implementations are better than some versions of Unix. A study by DH Brown Associates ranked four Linux versions better than the lowest ranked version of Unix. These Linux implementations were from SuSE, Red Hat, Caldera, and Turbolinux.

There are signs that Linux is gaining acceptance, at least in some areas. According to IT market consultants IDC, Linux had a 27 percent server OS market share last year--second only to the 41 percent of Windows.

Another IDC report shows a large adoption of Linux in the hosting market, topping US$3.5 billion in year 2000. Japanese mining giant Komatsu employs Linux as the foundation of a major online, spare-parts cataloging and tracking application. All these mean that Linux is running and powering a significant portion of the world's servers and expanding its reach.

From a functional standpoint, Linux is on par with Windows NT or Solaris. At the networking level, HSPcomplete from SWsoft is allowing mainframe class automation technology on these boxes, thus, increasing performance, availability, scalability and giving them more opportunities to gain large marketshare from the aforementioned growing market opportunity.

Will you get fired for buying Linux?

However, other enterprises--large, medium, and small--are reluctant to embrace anything but the familiar. A survey by Goldman Sachs indicates that Windows 2000 and Windows XP Professional dominate Fortune 1000 IT spending plans for the next year.

Linux, on the other hand, doesn't even register as a target. Although Linux may be inexpensive, it isn't totally free. Lack of spending plans indicate an obstacle to the growth of and an unwillingness to embrace Linux.

This is true even though Linux is widely acknowledged to be much more stable, to offer better application security, and not to suffer from the pricing practises of the competition.

What Linux doesn't offer are some enterprise features such as symmetric multiprocessing and the sense of familiarity. The former will be corrected over time, as large enterprise-focused companies, such as IBM and Compaq, work with the Linux community. The latter requires Linux to sit close to Windows. At one time, it was said no one got fired for buying IBM. The same is true today for Microsoft.

What is needed, then, are tools and implementations that take this into account. This is especially true in the small- and medium-sized enterprise arena, where IT staff is not traditionally large.

Change will now be required in the Linux versions themselves. For instance, ASPLinux 7.0 is based on Red Hat 7.0, with a modification to provide the Linux 2.4 kernel as an installation option. It also has been given a new installation program that can be run from Windows to partition and install ASPLinux as a dual-boot option on an existing Windows machine. Particularly for an SME, those types of features and flexibility are important.

Article contributed by SWsoft Pte Ltd with a few pointers from Alexey Kuznetsov who is recognized as one of the keepers of the kernel in Linux's inner circle. His day job is Chief Software Engineer, Networking, at SWsoft Pte Ltd.

Editorial standards