The use of Linux as an enterprise computing environment is growing rapidly, and there are strong business reasons behind this trend. Commercial Linux operating systems give companies a robust OS with powerful capabilities.

Linux enables IT shops to reduce hardware costs by using servers based on commodity processors and building-block architectures. It delivers lower total cost of ownership (TCO) compared to proprietary Unix and Windows environments. Linux users now have access to powerful third-party utility software. And Linux is constantly evolving and improving as a worldwide network of innovators in the open source community adds value to the Linux platform.

The challenge for companies that want to take advantage of these benefits is that Linux must become enterprise-ready - to reduce the risks or downsides before trusting their operations and their data, their most valuable asset, to this rapidly evolving open source operating system. They need to assure themselves that Linux can provide flexibility, availability, scalability, and interoperability in a multi-tier heterogeneous environment and, at the same time, keep their data protected, well managed, and always available.

To achieve that, companies must work with experienced vendors who can integrate Linux into their data centers and enable them to trust their data to Linux with confidence. A brief look at the recent history of enterprise computing is presented here.

The Historical Parallel: The Arrival of Unix

Then Unix emerged as a potential enterprise computing resource. In its way, it was revolutionary and it offered an appealing alternative to proprietary systems. Companies could now run applications on an open system, based on open standards, that could run on hardware from many vendors. IT departments were no longer shackled to the manufacturers of large mainframe systems. They could choose among competing vendors to drive down capital costs.

Sun Microsystems was highly successful in selling workstations based on Solaris, its version of Unix. However, companies with big data centers and large databases were understandably slow to embrace Unix. They were comfortable with mainframes. They had proven enterprise capability, provided high availability, and could easily scale to serve more users. They were robust, reliable, and redundant. IT managers were content to deal with a single vendor. Despite its promise of reduced TCO, Unix had a mountain of credibility to climb before it could make major inroads into corporate data centers.

What Made Unix Enterprise-Ready?

Two of these software products were instrumental in this change. One was the first commercially available journaling file system for the Unix environment. The other was a storage management system built around a volume manager. It allowed open systems to virtualize disks and disk subsystems, making storage management vendor independent. This development gave IT departments the flexibility to choose among storage vendors to reduce hardware costs.

Another major element in the creation of an enterprise-ready Unix was a backup server product that gave companies the highest level of data protection, enabling them to back up their data without bringing down their applications. Together these solutions played a significant role in enabling Unix to become a powerhouse in enterprise data centers.

Of course, these technologies were supplied in the Unix-based operating systems - AIX, Solaris, HPUX - included with the vendors' hardware. But the third-party tools provided much higher levels of reliability, functionality, performance, and failure protection - all dear to the hearts of IT managers.

Next Up: Making Linux Enterprise-Ready

Linux is making progress. It's now being adopted by corporate data centers to run nonmission-critical applications because of its substantially lower TCO, but it faces the same challenges that Unix faced: IT managers want to assure themselves that Linux is truly enterprise-ready, and that the high level of functionality they now enjoy is available with Linux before they move from open systems to open source. Most of these advanced features have never been available as defaults on Unix or any other operating system. However, they are available as third-party solutions for Unix environments, and IT staff members have developed considerable expertise and job security in their use.

Today, many of these third-party tools have been successfully ported to the Linux environment, where they will play the same role. This is a tremendous advantage for companies bringing open source into their heterogeneous environments. The toolset that a generation of IT departments is intimately familiar with, and comfortable with, will harden Linux and make it enterprise-capable just as it did with Unix. All of the experience built up over the last 15 years, and all the innovation and refinement of these tools, can now be applied to Linux. They will make the Linux OS much more robust and well rounded. People who are migrating from Unix to Linux or from Windows to Linux will continue to work with the tools they know well and use every day. The move to Linux becomes transparent in many ways.

Tools are becoming available that will make the move to Linux even simpler (see Figure 1). A new portability tool will free data from OS dependency. For example, a company that is running a third-party file system and volume manager on a Solaris-based system will no longer need to put data through a prolonged process to move it to Linux. Data can simply be moved to the new environment in a high-speed transformation that transfers terabytes in minutes. By the same token, data can also be moved quickly from Linux to many Unix-based operating systems, mitigating any associated risks.

Preserving the ROI of x86 Platforms

Over time, however, as data and the commodity platforms that manage it inevitably keep growing, the economies of scale reach the point of diminishing returns because of the increasing costs of managing hundreds of distributed processors. This problem is being solved today with automated server management and provisioning software that enables IT staff to manage hundreds of servers on the fly. It monitors the application, determines its requirements for processing horsepower, and assigns processors to it as needed.

As an example, assume you have 10 applications running on 100 servers. With the provisioning tool, none of the applications is tied to any specific server; they can run on any of the processors in the pool. An application may be running on 20 servers and experience a rise in user demand. The provisioning tool notes this and supplies the application with an additional 20 servers of processing power.

The provisioning tool can perform these tasks automatically on server clusters of any size. It also ensures that all the servers are current with the latest updates and patches. This efficient, automated management of the environment lets companies ensure that their TCO benefits - the reason they moved to Linux in the first place - do not go away.

Let's take this concept a step further. In the current Linux environment, each application is usually tied to a specific box because of data access requirements, which limits scalability to the confines of the box. In other words, to increase the workload for an application, IT departments normally have to increase the number of CPUs within the box so more CPUs can operate on the same dataset. But scaling up the box will probably be far more expensive than buying the same number of processors in two or more smaller boxes. This equation gets worse if you also invest in a stand-by box to give your scaled-up application high availability.

Instead consider a shared-data environment where a clustered file system can get you out of the box and let you scale out instead of up. It enables you to buy smaller, less-expensive boxes that can work together off a single virtual disk. It enables an application to be scaled out, across up to 16 nodes to increase processing capacity while working off a single dataset. It eliminates the need to create a symmetric multiprocessing (SMP) architecture to enable multiple processors to access a single dataset concurrently. It also inherently builds high availability since failure of any node keeps the application running on the remaining nodes. This enables users to scale out file servers, application servers, and even some popular database applications with confidence.

To Sum Up: A Hardened, Enterprise-Ready Linux