CNET también está disponible en español.

Ir a español

Don't show this again

Tech Industry

Commentary: Making the move to Linux

IT organizations should acquire experience with Linux as a potential long-term alternative to Microsoft Windows and Unix offerings.

With major vendors now devoting significant resources to commercialize Linux as a viable operating system for enterprise use, IT organizations should acquire experience with Linux as a potential long-term alternative to Microsoft Windows and Unix offerings.

Meta Group now recommends that large enterprises consider a limited number of Linux server deployments or pilot projects.

Currently, Linux is still a

See news story:
Chipmakers angle for Linux support
fringe player in the corporate world, with a small number of organizations using versions of the open-source operating system, typically on basic Web servers and mail servers. Linux is also popular for scientific/technical computing workstations and server farms.

However, there are two key reasons why Linux is likely to gain momentum in enterprise server implementations.

First, IBM and, to a lesser extent, Hewlett-Packard and other vendors are devoting resources to creating more robust versions of Linux that are suitable for widespread enterprise deployment. We expect Compaq Computer and Dell Computer will also accelerate Linux support. Intel is supporting the Linux bandwagon and may add specific hardware, firmware or software to optimize Linux. Although Linux offerings for enterprise customers are still too immature to justify widespread deployments, we believe "industrial strength" versions of the operating system will be available in two to three years.

IBM in particular is investing significant money and pushing hard to make Linux work. The Linux start-up vendors haven't succeeded because they had no way to make real money. Traditional vendors such as IBM and Intel will make money by using Linux to sell more hardware and services.

Second, aggressive pricing and licensing changes for use of Microsoft Windows 2000 on servers are driving customers and competitors alike to seek an alternative. In the past six months, we have encountered a strong wave of user animosity toward Microsoft in the wake of server pricing changes to its Client Access License. However, we believe Linux-based software and services will increase in cost over time to a level roughly about 20 percent less than the pricing of Microsoft's Windows NT and Windows 2000. Current Linux software pricing is generally too low to be sustainable.

Costs and benefits
Linux's initial "sex appeal" as open-source software is irrelevant from a corporate perspective. Enterprise customers welcome the investment by IBM and others to create sturdy and easily supported and managed versions of Linux, even though this means that Linux will inevitably fragment into different (proprietary) versions.

Linux is typically not a low-cost alternative when viewed from a total-cost-of-ownership perspective, because it costs more for organizations to support it. We view IBM's commitment to Linux as a major factor in making the operating system a viable alternative that can be reliably and repeatedly installed, operated and managed on enterprise infrastructure.

Microsoft's recent licensing and pricing changes for Windows server deployments make perfect sense--from the point of view of a Microsoft stockholder looking for maximum profits over the next several years. The company is exploiting its commanding market position to extract as much money as possible from its customers.

However, Microsoft's aggressive pricing is driving the momentum of Linux. Enterprise customers are angry about the larger bite that Microsoft is now taking out of their budgets with its latest Client Access License pricing. If a reliable version of Linux can be used on Web servers at a lower total cost of ownership, it will find enterprise customers ready to listen. Whether they truly forsake Windows or simply use Linux as a hedge and negotiating tool against Microsoft's pricing demands, consumers want to have some choice and leverage. On the desktop, of course, aside from technical workstations, there is no vendor support for Linux deployments.

Gains by Linux will also come at the expense of Unix operating systems (such as Solaris, AIX, HP-UX). As Intel increasingly gains 64-bit technological equivalency with proprietary Unix chip offerings such as Sparc at lower prices, Linux on Intel will change the pricing equation at the expense of Sun Microsystems. Meanwhile, IBM and HP, with their Linux development efforts, are in effect hastening the eventual obsolescence of their Unix systems. This reinforces the more powerful trend of "n-tier" application architecture driving Unix toward the backend, database management system tier and eventual legacy status.

The position of Linux now is similar to that of Unix in the early 1980s. Throughout the 1980s, Unix was viewed as "snake oil"--not really a viable commercial system. For a long time, it sort of percolated through the industry, and then with enough persistence from enough people it finally broke through. Our research indicates that Linux will mature faster than Windows NT and 2000, which in turn matured faster than Unix.

Intel and IBM are now doing to Sun what Sun did to Digital Equipment in the mid-1980s. They are making an operating system with a tight hardware-software integration (such as DEC VAX in the 1980s, and Sun Solaris now) much less relevant, and ultimately turning the server market into a commodity (virtually entirely Intel chip-based) market. Ultimately, Linux will get fragmented and vendor-specific like Unix, but for now it has established a new commodity-like dynamic that favors high-volume, low-cost producers. With Intel's new generation of chips, almost any manufacturer can produce a high-performance box at a reasonable cost. Sun now has a two-front war to fight: Microsoft on one side, and the Intel/IBM/Linux camp on the other.

Not quite ready for prime time
Meanwhile, we are seeing a polarization and shakeout of the Linux community, with Red Hat on one side battling Caldera International on the other. In the "Lintel" world, Linux on Power or mainframe systems is a niche, but on single-function servers or application appliances, Linux is becoming viable. However, Linux is still two to three years from being a database engine or usable in a load-balanced cluster.

One factor in Microsoft's favor is the costs for enterprises switching from Windows server implementations to Linux. It will be much easier for a Unix shop to deploy Linux and get positive returns than for a Microsoft shop. Most Unix software has been, or will be, ported to Linux over time, so a Unix shop switching to Linux doesn't have to worry about switching database and application server platforms.

For a Microsoft shop, switching to Linux will require retooling--products, skills, personnel, procedures--not only the operational environment, but also the entire development environment. This will be a huge investment, and unless the bottom drops out of Java application-server pricing, software licensing will probably still be more expensive for Linux/Java than for Microsoft, even with Microsoft price-gouging.

Meta Group believes that businesses should gain in-house expertise and experience with Linux deployments on Web servers, mail, file and print servers, and--eventually, as more software becomes available--Java application servers as a defensive play against predatory pricing by Microsoft and to exploit future Linux offerings by IBM and other major vendors.

Although widespread Linux deployments are not yet feasible, we recommend that companies pursue smaller-scale Linux implementations, pilot projects, and development work by future-technology groups.

Exploring Linux--especially on the server side--should be part of an organization's IT investment portfolio. It may cost a little bit more to support Linux now, but this should be viewed as an insurance policy to hedge against future Microsoft incursions into the corporate wallet.

Visit for more analysis of key IT and e-business issues.

Entire contents, Copyright © 2001 Meta Group, Inc. All rights reserved.