Over the past couple of years, Linux has transitioned from the IT fringes to the technology mainstream. With giants such as IBM, HP/Compaq, and Computer Associates (CA) now firmly on the Linux bandwagon, Linux has surpassed UNIX to move into the number-two position in terms of new server operating systems.

According to the latest research figures, this rate of expansion will continue to accelerate. Gartner predicts that the Linux market will surpass $9 billion in revenue within four years. International Data Corp (IDC) concurs, anticipating a growth rate of 28.2% annually for Linux until 2006.

As the number of Linux applications multiplies, its penetration into the corporate world continues to gather steam. Just a few years ago, Linux enterprise involvement was restricted largely to services such as file and print sharing or simple Web serving. While these functions continue to be important, Linux is increasingly being utilized for enterprise-scale e-mail, firewalls, and database servers, and for a wide range of network and network management services.

Linux is also up to date in the technical sense, and it is playing a vital role within organizations in the deployment of all sorts of next-generation technology. Organizations developing sophisticated new applications based on emerging Web services or grid computing standards are finding Linux to be an ideal server platform. As these new technologies become more commonplace, Linux stands to have an increasingly important role.

Further, Linux has distinct scalability and portability advantages compared to other architectures. Applications developed for Linux can scale from small Intel commodity boxes up to the largest mainframes with very little effort. CA has validated this with the ability to take applications originally developed for Linux on the Intel architecture and have them run on mainframe Linux with very little effort. CA's experience - spanning millions of lines of code and dozens of products - demonstrates that Linux is robust enough to provide the kind of portability across architectures that was only dreamt of in the past.

The scalability and low cost of creating applications that function across multiple architectures is an important advantage that will help Linux continue to gain strength in the server marketplace. Instead of making platform-hosting decisions based on the relatively small "sweet spot" offered by a single platform and architecture, Linux makes it possible to mix and match according to the needs of the organization. Companies can pick the platform - from Intel boxes up to mainframe - based on the number of users, as well as price/performance, market conditions, reliability, and security issues.

An example of this at CA is an important in-house HR application that operates using an Ingres relational database. For 10 months of the year, it is hosted on a medium-sized Sun server that copes well with routine demands. But with 16,000 employees worldwide, enormous traffic is generated during the annual performance review cycle. During that period, CA moves the backing Ingres database to a mainframe Linux environment to handle the peak load. Except for improved response times, users are generally unaware of the difference, and CA doesn't have to dedicate otherwise unused capacity to the application year-round.

Mainframe Linux
Without a doubt, enterprise Linux deployments to date have paralleled the ascendancy of commodity Intel server boxes. The rise of the Internet and the sweeping adoption of distributed applications over the past decade have generated an explosion in the number of servers IT must manage. Due to its sheer cost effectiveness, Linux has successfully invaded this market and is steadily rising to dominance within it.

The rapid increase in distributed servers seemed more attractive than a centralized mainframe-based computing model - at first. But in some organizations, server proliferation has begun to spin out of control. These days, it's quite common for organizations to be running hundreds or thousands of Intel and UNIX servers, many of them single-application boxes. It's also common for companies to have no idea exactly how many servers they have, or what applications rely on which combinations of servers.

This proliferation of commodity servers has proven far more costly than many previously predicted. Though smaller, scaled-out systems are less expensive than mainframes, IT needs many more of them to support the business. That translates into high costs, more exposure to potential security breaches, and a greater management burden.

In response, server consolidation continues to blossom. Gartner says 70% of its enterprise clients are now trying to consolidate, compared to only 33% three years ago. Many of these consolidation projects involve Linux, and vendors like IBM and CA are heavily involved in the process.

It is possible, for example, to consolidate UNIX and Linux servers onto a mainframe such as the IBM zSeries. Under this model, you can transform a single mainframe into hundreds of virtual servers, each running individual copies of Linux.

One midwest hospital chain, for example, had 40 small Intel servers hosting e-mail, Web server, directory, and various health care applications. It consolidated them onto one IBM eServer z900 running multiple instances of Linux. Its health care claim-processing system originally spanned multiple servers. As the amount of claims on this system grew from $800 million to almost $6 billion over four years, it became a constant source of trouble for IT. It took integration onto the mainframe/Linux environment to bring about the needed improvements in efficiency. That organization now processes 350,000 claims daily - 12 per second. The IT manager reports that he can create a new virtual server in two minutes compared to a day or more per server for Intel boxes. Other advantages cited include a 50% cost reduction per server, a performance increase of 3 to 1, and a mean time to failure that's gone from 6 years to more than 30 years.

With the successful integration of so many infrastructure servers, the hospital chain is currently involved in the next step - application integration. This includes a PeopleSoft migration as well as database consolidation from SQL Server to IBM's DB2 Universal Database. Once again, Linux is the operating system of choice.

While the mainframe Linux model may not be right for every situation, it is certainly growing in popularity in the enterprise. So when does consolidation make the most sense?

• When multiple distributed infrastructure servers need to handle tasks such as file and print, Web, and e-mail
• When consolidation of multiple databases is needed
• When there is a need to manage more with limited resources
• When there are sites with excess mainframe capacity
• When application peak traffic periods are complementary
• When scalability and performance needs are extreme
• When applications can benefit from close proximity to existing mainframe data
• When more space is needed in the data center or where environmental costs are high
• When the volume of distributed servers becomes too great to keep up with routine maintenance, updates, etc.

Obviously, each platform has its own sweet spot in terms of the number of users it can comfortably support and the amount of complexity it can manage. It wouldn't make sense to host one small application on a mainframe or to attempt to cobble together a huge application on a few distributed Intel boxes. Therefore, it is up to IT to mix and match according to needs. But clearly, mainframe Linux has a major role to play as the server consolidation trend takes hold. And as the platform has roughly the same management costs on Intel as it does on mainframe, look for Linux to become the OS of choice in server consolidation projects over the coming years.

Blade Server Linux
Another form of consolidation that has Linux in the forefront is blade servers. Instead of moving from commodity boxes to the mainframe, though, blade servers house multiple boards on one chassis, thereby cutting the power, cooling, and hardware costs of computing. IBM, Intel, Dell, and HP, to name a few, are investing heavily in blade servers.

Market research firm IDC forecasts the market for blade servers will rise to $3.7 billion in 2006, from $133 million this year. In all likelihood, Linux will become the preferred OS. HP, for example, launched a range of blade server offerings based on Linux. Because Linux is lightweight, has lower memory requirements, and makes better use of CPU power than alternative operating systems, it is ideal for blade servers, which require flexibility and scalability due to heat restrictions.

AOL, for instance, recently announced that it is testing blade servers with a view to widespread adoption. With an average of 2.5 million subscribers active on AOL at any given time, its 800 RISC-based UNIX servers are kept busy managing log-ons and user-name and password verifications, as well as parental-control activations. As many of these servers are coming up for retirement, blade servers running Linux are likely to take over as a more efficient way to handle the load. So far, AOL has installed 56 IBM BladeCenter servers. These have 14 "server-on-a-board" systems per chassis. Each runs Red Hat Linux 7.3. Once the pilot demonstrates the cost, scalability, and performance advantages anticipated, AOL will replace all 800 UNIX systems with Linux-based blade servers.

Desktop Linux
For many years, analysts have announced Linux as the rising force in the server marketplace. With mainframe and blade server Linux systems now adding to the established trend of widespread Linux adoption on commodity Intel boxes, that prediction has come true even more forcefully than many expected. Those same analysts, however, have always maintained that Linux would never become a major player on the desktop - until recently.

Over the past six months, in fact, there has been a significant shift in the willingness of enterprises to investigate the large-scale commercial usage of Linux-based PCs. Faced with harsher economic times, many businesses are rethinking their approach to the desktop and have become amenable to new strategies.

Many modern business applications, after all, are browser based. Rather than a high-powered and expensive desktop loaded with proprietary software, some are evaluating a thin-client desktop model based on open source elements. As desktops come up for replacement, some companies are looking to cut costs. Instead of buying the latest desktop hardware/software, they see the value of harnessing Linux to gain several more years of use out of existing hardware.

Companies like Merrill Lynch, for example, are looking into this approach as reported widely in the media. Since Merrill Lynch began working with open source software, it has significantly reduced hardware and software costs. Currently the company is investigating Linux on the desktop with the initial goal of 20% of its desktops using open source products.

Embedded Linux
In the area of embedded applications, Linux has risen to prominence with a minimum of fanfare. As these systems are largely invisible to users, they receive scant attention in the media. Yet here, too, Linux has quietly conquered the market.

Embedded Linux involves porting the Linux kernel to run on a particular CPU and board, which will be put into an embedded device. For the most part, the APIs and kernel are the same for embedded Linux as for desktop Linux.

Many of today's common computer devices, such as wireless hardware, routers, switches, and other networking hardware, all use Linux to run embedded systems. High-profile consumer devices like the Sony PlayStation and TiVo utilize Linux-based embedded systems. A number of major automakers such as DaimlerChrysler, BMW, and Ford are also using Linux in electronic products. Why? Instead of designing application-specific hardware, these automakers are now beginning to get behind cross-industry standardization using a Linux kernel to plug any device into it. The cost advantages could be staggering.

BMW, for instance, has developed a wireless environment to collect serviceability data on each vehicle, which can be relayed to the dealer. Once that system is up and running, the dealer can monitor the smooth operation of sold vehicles. The dealer can then tell a driver to bring the car in for a check-up based on performance criteria falling outside preset thresholds. Interestingly, such systems open up a whole new area of risk management and security, which CA is helping to resolve. Before you relay any vehicle data wirelessly, you have to ensure user privacy and overall system security. Only when that is well in hand will it be possible to broadly use such a system, or introduce innovative approaches to vehicle security such as biometrics, i.e., validating user IDs in order to start a car with a thumb scan.

Linux and Grid Computing
Yet another area of the IT landscape being successfully invaded by Linux is grid computing. Simply put, grid computing technology enables access to applications and data, processing power, storage capacity, and a vast array of other computing resources. It involves clusters of servers that are joined together over the Internet and uses standard protocols, and it is increasingly harnessing open technologies such as Linux.

Among other things, grid computing allows:

• Coordination in or between physically dispersed virtual organizations.
• Availability of computers, data, software, storage, and other resources.
• Controlled access by resource providers and consumers, which defines who can share, what is shared, and which conditions allow sharing.

The interoperability of this network transforms the grid of servers into a single, large virtual computer for the end user. How this model will evolve remains to be seen. It may eventually morph into an entirely new computing platform, or it could grow into an alternative to today's lower-level platforms. Whatever happens, though, Linux is already widely used in this development and will lead the way. Already, IBM and Ford have invested heavily in the development of economical commercial uses for grids, and CA is working to provide the management framework that viable grid computing demands.

Perception Trouble
While gazing into the crystal ball is never an easy proposition, it seems certain that Linux will continue to steadily increase its market share in the server, desktop, and embedded marketplaces. It remains to be seen whether Linux will rise to the occasion and defeat Microsoft in the desktop marketplace, but its dominance in the server space is assured.

Why? The tide appears to have turned with regard to corporate perception of Linux. Up until about two years ago, most executives were reticent toward Linux. Their apprehensiveness was routed in ongoing support and security concerns about the open source movement as a whole. Some feared that any hacker out there could gain access to Linux systems and that they were better protected using proprietary systems. But as the recent spate of security and virus alerts has demonstrated, proprietary software is, if anything, more open to attack that open source products.

More and more these days, CTOs are vocalizing that once they become familiar with the open source model and experience how it works, they realize that more eyes on the source code means greater security in the long run. The bottom line is that more people are actively looking out for the security of your systems.

That's why Linux is experiencing more interest from clients than ever before. In 2001, Linux came up less than 10% of the time during client interviews. Today, it comes up two-thirds of the time. That percentage is increasing markedly from month to month, and it seems only a matter of time before every major software and hardware vendor is forced to adopt a full-blown Linux strategy in order to remain competitive.

Linux Creates Reorganization at CA
In the mid to late '90s CA decided it was worth placing a strategic bet on Linux based on early customer adaptors. For most of its 27-year history, CA has concentrated on the development of products that break down the barriers between different platforms - either to manage the heterogeneous enterprise or to act as middleware between complex systems. CA products support everything from IBM mainframes running z/OS and OS/390 down to PDAs running Pocket Windows. In keeping with this tradition, CA is now firmly behind Linux with over 60 CA Linux applications now available for the distributed and mainframe environment.

What initially caught CA's attention was the fact that a range of efficient tools didn't adequately support Linux. In particular, Linux servers didn't integrate well in an enterprise environment with equipment running on other platforms. With more and more clients deploying Linux, it was up to us to come up with the software.

CA released its first Linux products in April 1999, management agents for the Unicenter TNG management package, to be followed up later that year with the complete network and systems management package. Since then, CA has steadily continued to port products over to the Linux platform. Each is compatible with popular versions of the Caldera, Red Hat, SuSE, and TurboLinux versions of the Linux operating system.

For enterprise management, CA offers modules for Unicenter covering service-level management, job scheduling, Web server, WebSphere, and cluster server management as well as network and systems management for devices, applications, and databases running on Linux. The BrightStor storage management line also has products specifically designed to back up and restore Linux systems, and CA enterprise back-up software supports Linux in a heterogeneous environment.

In addition, CA offers a variety of security management applications covering virus protection, policy compliance, security monitoring, and access control either in a Linux environment or for Linux boxes operating in a mixed environment.

To support this major investment in Linux, CA has established a dedicated Linux Technology Group that spans all of its major brands:

1. Enterprise management (Unicenter)
2. Storage (BrightStor)
3. Security (eTrust)
4. Portal and business intelligence (CleverPath)
5. Database management and application development (Advantage/Jasmine)
6. Software life cycle management (AllFusion)

Rather than being one product group within the company as a whole, the Linux team operates at a strategic level. This has resulted in the formalization of CA's Linux efforts and the integration of the operating system into all of CA's product lines.

The Linux Technology Group has three main aims: greater optimization of Linux in terms of reliability, availability, and performance; securing and preserving data, applications, and systems to minimize risk; and integration of Linux into the cross-platform environment in order to allow seamless integration with the heterogeneous nature of the modern enterprise.