Dell Insight

(February 2003)

Download article (430K)

Download Issue (6.13 MB)

Dell Insight: These days, we hear a lot of talk about "open systems." What are open systems?

Stacey Quandt: The term open systems refers to efforts of the Open Systems Foundation, now known as The Open Group, to create an open operating system (OS) environment. The activities of OSF and the UNIX^®  vendors that joined it promoted open systems as an alternative to the alliance between AT&T and Sun Microsystems to deliver a single unified UNIX OS. The benefit of the open systems movement was that a single UNIX vendor no longer could monopolize UNIX implementations. The disadvantage, of course, is that vendors developed competing proprietary versions of UNIX. This situation created a problem for software vendors because they needed to port applications to each different UNIX offering. Interoperability often was a problem.

DI: That's where "open source" enters the picture. Can you explain the distinction between "open source" and "open system"?

SQ: The basic distinction is that open source means that anyone has access to the source code, whereas an open system does not provide this level of transparency. For example, the Linux^®  operating system allows anyone to view, modify, and distribute the source code. Users can download the Linux OS from the Internet or purchase a distribution from Red Hat or other Linux distribution providers. You also can replicate Linux across multiple servers at a lower cost than you could with other environments. Enterprise-class versions of Linux from Red Hat, Sun, and UnitedLinux narrow the performance gap between Linux and UNIX. Although an enterprise-class version of Linux will cost more than a mainstream Linux distribution, it is coupled with service and support programs.

Another important distinction between "open systems" and "open source" is licensing. Open-source solutions are available under a variety of licenses so that anyone can tweak the source code, whereas providers of open systems do not offer open-source licenses to access their source code.

DI: Other than up-front cost savings, what potential business benefits can Linux deliver to enterprises?

SQ: The main business benefit of Linux is its ability to reduce IT costs. Another potential business benefit is that Linux can deliver up to three to five times the performance of a UNIX/RISC system at half the cost. The performance and cost benefits of Linux have made it a volume operating system second only to Windows on industry-standard Intel^®  x86-based servers. In comparison, a UNIX solution is often more expensive than Linux because it is designed to run primarily on microprocessor architectures that have high margins but limited market share. This distinction has led Linux to become a flexible choice for customers—critical to its rise as a mainstream OS.

Due to a high degree of interoperability, Linux can succeed where UNIX has failed. This aspect has driven independent software vendors (ISVs) to port more applications to Linux. The advances in compatibility among Linux distribution providers enable software applications to run on any compliant Linux system, making Linux a flexible tool for IT managers. The flexibility of Linux gives customers the ability to choose among a variety of hardware vendors, and the multiplicity of Linux distributions means that users are not locked into buying an OS from a single provider as they are with Microsoft^®  Windows. Like any standard technology, Linux is becoming a commodity. As this commoditisation happens, increased competition among vendors gives end users the power of choice.

DI: Compared to other operating systems, is the number of mission-critical Linux deployments still relatively low? Is this because concerns still hinder the advancement of Linux in the enterprise and, if so, what are vendors doing to address those concerns?

SQ: First, some technical issues still hinder the advancement of Linux in the enterprise. What I mean is that Linux has come a long way in 10 years of development, but it is still maturing. Linux is certainly a contender in midrange server deployments and can scale well up to an x86-based 8-way symmetric multiprocessing (SMP) system, but the Linux community and industry vendors are working to advance the performance capabilities of Linux up to a 64-way SMP.

The second issue that hinders the advancement of Linux is the comfort level of end users. Some UNIX and Windows^®  shops remain skeptical of swapping out a legacy system in favour of Linux or adding Linux as one more operating system to support. The challenge for some C-level executives is in understanding the performance capabilities and economic benefits of Linux. As Linux matures, this perception will change. Of course, the best way for CIOs and system architects to overcome their concerns is to benchmark the scalability of Linux in a test environment to see that it can support a number of enterprise workloads.

Linux has crossed the technology adoption chasm in the context of Web servers—Apache, for example—application servers, and file and print services deployment. A similar process for using databases on Linux is underway.

Today, Linux is becoming a core element in both workload consolidation and support for the distributed enterprise, and more enterprises are considering a Linux platform. The availability of Linux service and support from software and hardware vendors has made Linux a contender in the enterprise and helps reduce the reluctance of IT decision makers. Also, benchmark tests such as TPC-C, TPC-H, SPECweb99, and others can provide convincing arguments about the capabilities of Linux.

Perhaps the most vital validation of Linux achieving mainstream acceptance is the number of customer references from companies such as the FAA, IMAX Corp., Merrill Lynch, Reuters, Morgan Stanley, China Telecom, Amazon.com, Danish National Railway Agency, and many others. Customers deploy Linux in an effort to reduce IT spending, leading more executives to investigate the possibility of leveraging Linux within their own enterprises.

DI: But Linux isn't the right cost-cutting measure for every enterprise, right?

SQ: That's absolutely correct. I often hear executives say, "Oh, we should use Linux because of the cost savings that everyone else is experiencing." This perception does not always lead to the right answer. For example, if users are running applications on Windows and want to migrate their IT infrastructures to Linux, that exercise is often non-trivial. This is especially true if users need to port their own custom in-house applications from Windows to Linux. The initial pain of migration can pay off later—especially in the context of server consolidation and associated savings on software licenses.

However, keep in mind that it is often more difficult to port software from Windows to Linux than from UNIX to Linux. In fact, a great deal of the cost savings highlighted today are specific to a UNIX-to-Linux migration. Often, the most significant savings occur with hardware cost—moving from a RISC-based to a standards-based x86 Intel architecture.

DI: We have seen some incredible advances for Linux in the enterprise during the last few years. What major changes do you foresee during 2003?

SQ: The growing stability of the Linux 2.4 kernel coupled with high availability, robustness, and reliability will drive more organizations to evaluate Linux for mission-critical workloads. This growing maturity will also lead to more scrutiny of the comparative features and functions of Linux, UNIX, and Windows in terms of scalability, systems management, security, and Web application services. The anticipated release of the Linux 2.6 kernel in 2003 will enable even greater levels of scalability and performance. The maturation of Linux coupled with ongoing Microsoft Windows security issues will lead users to compare the benefits and disadvantages of open-source and closed-source operating systems.

The momentum behind commodity Linux deployments is seeping into the telecommunications industry. Telecommunications manufacturers are beginning to consider Linux on Intel processor-based servers as an alternative to a proprietary OS.

DI: Beyond 2003, where do you see Linux in five years? How deeply will it penetrate into the enterprise data centre?

SQ: Since its inception 10 years ago, Linux has matured into a strategic tool for CTOs, CIOs, and system architects to reduce IT spending. Within the next five years, we will see Linux evolve into a core component of enterprise data centres and support application workloads that are primarily supported by UNIX today.

Cost constraints and the need to demonstrate return on investment (ROI) will continue to play a major role over the next five years as more companies migrate their infrastructures. Based on server revenues, Linux distribution providers accounted for a $1.5 billion market in 2000, a $2.5 billion market in 2002, and will grow to a $15 billion market by 2007. If current market dynamics do not subside within the next three years, Linux will emerge as the dominant operating system based on unit server shipments compared to Windows and UNIX. In addition, as next-generation hardware supports up to a 64-way SMP system, the gap between Linux and UNIX will narrow.

Already, Linux is a contender for database management system deployment and the Linux platform is an attractive infrastructure on which to deploy proprietary relational databases. Although enterprises today use Linux to power database programs, they generally do not use Linux for data warehousing. Data warehousing might be an area of growth for Linux within the next five years.

Also, Asia Pacific and EMEA very likely will experience a greater adoption rate of Linux compared to North America. Currently, Linux adoption is growing exponentially in geographies such as India, China, and Brazil. Over the next five years, we expect Linux will overtake Windows in unit server shipments to those regions.

DI: What about the future of Linux in the high-performance computing (HPC) cluster market segment?

SQ: In the HPC cluster segment, Linux already can support extremely parallel workloads such as reservoir analysis, aerospace engine design, risk analysis, nuclear simulation, drug discovery, and crash analysis. Customers can take advantage of the synergies between the economics of Dell as a low-cost hardware provider of industry-standard servers coupled with Linux to support highly parallel systems.

Overall, the growing number of Linux supercomputers is evidence of a significant shift for both the academic and commercial communities. Increasingly, users in vertical market segments such as oil and gas, automotive, life sciences, research, and financial services are adopting Linux supercomputers. The implication for Dell is that teraflop-level Linux machines are becoming the norm, and Linux commodity clusters are redefining high-performance computing. Linux clusters represent a rapidly evolving market segment, with further developments anticipated in cluster-based algorithms, cluster hardware, system configuration, and cluster management. Dell's focus on Linux-powered blade servers is yet another point in the evolution of supercomputing.

DI: Can you forecast any trends for deploying Linux on the desktop?

SQ: Overall, the adoption of the Linux desktop will be incremental—especially in North America—as a replacement platform for office productivity applications. China, Belgium, Peru, Spain, and other countries appear more willing to embrace a Linux desktop.

Microsoft continues to dominate the office productivity suite market with approximately 95 percent market share, according to a recent survey of Giga clients. However, Microsoft continues to struggle to convince its customers to upgrade to newer versions of its Office suite. This fact signifies an opportunity for users to consider Linux. A Linux desktop is likely to be a realistic alternative within vertical markets such as government, retail, financial services, and education. Application segments in which Linux desktops show promise are point-of-sale systems and call centres.

We might be far from the day when the average consumer boots up a Linux-based home PC. But the growing maturity of Linux and its success in existing server implementations should leave no question that Linux is here to stay and will only get better with time.

Stacey Quandt is an industry analyst focusing on Linux and open source software. Her current research topics include Linux distributions, the Linux server market, and open source issues, trends, and best practices. Stacey has advised many North American, European, and Latin American clients on adopting Linux within their enterprises and consulted with companies launching Linux-based products and services.