Columns
The move to open source
Over the course of the past several years, I’ve written about open-source software in general and Linux in particular. My main question was: “Is Linux ready for the enterprise?” In the last few months, there has been a surge in reports of enterprises adopting Linux. Typically, the reason given was lower cost or increased security. But is this apparent surge an anomaly or a trend? In these economic times, a desire to reduce costs is understandable. The “free” nature of open-source software may appeal to many firms as one way to reduce costs. Meanwhile, the impression that Linux is more secure than Windows has been fostered by a never-ending set of security patches for the various flavors of Windows. Whether this impression is true or not remains to be seen. Linux is not invulnerable.
In their study “Fear the Penguin,” analysts at Goldman Sachs stated, “Linux will emerge as the dominant operating system in corporate data centers.” Their basic rationale is that Linux on Intel-based services is a lower cost alternative to premium Unix systems running on RISC processors. They also believe that the performance of these Linux/Intel systems is adequate for the needs of many enterprises. For years, the Linux, Apache, mySQL, Perl/PHP/Python (LAMP) combination has been a popular approach to the Web. Linux is also easy to configure for file, print and e-mail servers. An important conclusion of the study is that the growth of Linux in the enterprise will not stop there. Goldman Sachs believes Linux will come to dominate on the higher-end servers of the data center as well.
Analysts at Forrester Research came to a similar conclusion. They believe cost pressures will make Linux a compelling purchase for the data center. Although they also see Linux as dominant, Forrester analysts believe that there will always be a portion of the enterprise that needs the higher capabilities and performance of RISC-based systems. They believe Linux will have an even greater impact by hindering Windows’ entrance into the data center. Key to this is the easier migration path Linux will provide for applications that currently reside on Unix. It’s interesting to note that both studies believe that Linux displaces Unix. But is this the case? A recent study by Evans Data Corp. finds that most firms moving to Linux are doing so from Windows, not Unix, platforms. If this trend continues, it paints an even bleaker picture for Windows’ hope in the enterprise.
Although the number of corporations using Linux is still a minority, it is a fast-growing group. According to IDC, in Q4/2002, sales of servers pre-loaded with Linux software rose 41% from the year before. Overall, noted IDC analysts, “Linux servers represented some US$600 million in sales, compared with $5 billion in outlays for machines running proprietary Unix.” This is consistent with what Hewlett-Packard (HP) is seeing with Linux. HP recently signed an agreement to support Red Hat Linux in the enterprise. In Q4/2002, HP Linux server sales grew to $80.2 million from $44.3 million the year before. In January, HP announced that it had pulled in $2 billion in Linux-related revenue in 2002.
This is fine as far as predictions and sales of servers and software go. But who is actually using Linux? Is it making the inroads predicted? There are, indeed, some well-known companies moving in this direction. Amazon has announced its move from Solaris to Linux running on HP, and Oracle said it is switching its systems from Unix to Linux. And E-Trade noted that it would be moving its entire architecture to IBM Intel servers running Linux.
I’ve looked at the role of open-source software and Linux in the financial market, and Linux is making inroads. Morgan Stanley’s IT department was tasked with reducing the cost of its computing infrastructure, and the Linux/Intel combination allowed them to develop a low-cost infrastructure. JP Morgan Chase has also announced that it is moving back-end systems to Linux. And Banco do Brasil has said it will consolidate its European back-office banking systems onto IBM systems running Linux.
Others have found Linux an appropriate platform for running applications. Recently, Reuters released a Linux version of its Reuters Market Data System used by financial institutions to integrate real-time data and news used in trading decisions. In a similar vein, Credit Suisse First Boston (CSFB) rolled out its global trading system, Agora, on Linux. There was a 20-times increase in system performance by moving from Unix and RISC to the Linux/Intel platform. According to CSFB, its trading desk made an extra $20 million because of increased trade flow. Drivers for this move were total cost of ownership and Linux’s ability to support innovation.
But what about other open-source applications? Do the many advantages often claimed for open-source software apply in today’s environment? We need to be clear that everything that runs on open-source software is not open-source software. Oracle is a perfect example of this. Some in the open-source community believe that everything needs to be open-source software. This is a political stance that makes little sense for business. Where open-source software has an immediate effect on cost savings and total cost of ownership -- fine. But most companies use a wide variety of proprietary software that they are not about to abandon. This is a situation where open-source software can work hand-in-hand with proprietary software. Often an open-source system such as Linux provides the platform for proprietary software. As more proprietary software becomes available on Linux, the more likely it is that companies will consider Linux for their infrastructure. Currently, this combination is needed because there is a dearth of end-user and business apps available as open-source software.
The LAMP collection of open-source software is representative of many open-source applications. A notable exception is the OpenOffice system. Many view OpenOffice as an open-source replacement for Microsoft Office. In addition, many believe it is found only on Linux. Strangely enough, a survey found that 63% of OpenOffice users were running it on Windows. This appears to be another form of cross-pollination between open-source and proprietary software. This points out that using a particular piece of open-source software does not commit a business to using other open-source software. For the business, cost savings and total cost of ownership can help to determine where open-source software fits and where it doesn’t.
Businesses have little interest in the fervor and jingoism that at times surround open-source software. We need to be clear that there are advantages and disadvantages to open source. Consider licensing. While current licenses may be fine for users of the software, there are often terms that keep others from adopting code and including it in their own because the license requires that their code also become open source. This is a nonstarter for vendors and others who would adopt open-source software that embeds standards or best practices. Also, because of the way open-source projects are typically done, there have been a significant number of failures. This is to be expected given the ad hoc nature of most open-source projects. Many projects are birthed and die quickly. I’m very glad we have notable exceptions. I hope there are many more, but that is unlikely unless the model for open-source software development changes.
This does raise the issue of how business application open-source development should be done. The developer-centric approach will not work. A sustainable approach, focused on the needs of the end user, is required for business apps. This approach must also encompass the mechanisms to provide support for end users and organizations.
For business, it is clear that immediate cost savings and total cost of ownership will continue to be drivers in moving to any new technology. Political diatribes and jingoism are of no interest to business. What is of interest is the mounting evidence of the value open-source software can bring to a business. So what should a business do? Take an approach to open-source software that seeks value for the business. Look for combinations of open-source and proprietary software that reduce your immediate costs and improve your total cost of ownership. Encourage maturation of the open-source industry by insisting on quality support. Finally, share with others what you are doing.