After dropping various thinly veiled hints here and there in articles for the last few years, I finally decided it was time to come out of the wiring closet. This issue, we will dive deep and examine the Gulf Stream of Internet undercurrents - Open Source Software. I've been a fan of the open source movement for a number of years. Given the relatively conservative nature of the Federal government, though, you just do not see a lot of government systems whose development and maintenance depends primarily on the goodwill of their user communities.
However, due to a number of factors both within and outside the open source community, I believe the open source software movement is poised for explosive growth over the next10 years and, if you'll pardon a sporting metaphor, ready for prime time. Yes, I know that's what stock market analysts said about dot-com initial public offerings two years ago. But this is different.
First, most of the software we'll look at here is, for the most part, free. You do not have to send any money to download the Linux or FreeBSD operating systems, the StarOffice application suite, or the Apache Web server. There are also a wide variety of free e-mail clients available for open source staple Sendmail. If you have a spare computer lying around you can download all sorts of things and start tinkering without a single contract.
Second, unlike much proprietary commercial software, you can download the uncompiled source code for all these applications without signing a non-disclosure agreement. Computer security experts can peer into the darkest corners of the software looking for flaws, bugs and loopholes. Even better, they can fix and redistribute the source code on the spot, which often happens within an hour of detection. Contrast this with how long it even takes to find out that proprietary software has a bug, let alone fix it.
Finally, the reason the community approach works is not because of altruism, but self-interest. The same people who develop and maintain open source products use them to run their own critical applications. Therefore, they have a very strong self-interest in keeping them secure, economical, and effective. Because of the open nature of the source code, it's virtually impossible for bugs or flaws to hide for long. Too many people whose operations depend on the code rigorously examine the code every day, even long after a version is released.
It is the 2002 equivalent of people hanging around the garage tuning their hot rods for the Friday night drag races. In this article we'll take a brief look at open source, some of its flagship products, the feasibility of replacing proprietary systems with open source, and some associated cost of ownership issues. First, let's see how Zippy is dealing with impending fatherhood.
Open Source Maternity
For those of you who might have missed last issue's LPG (Lazy Person's Guide) installment, Zippy and Zippette are expecting a blessed event sometime this summer. The big news lately is that it will not be just one baby, but two, which really makes sense considering Zippette is the closest thing to a pure binary being a carbon-based life form can aspire to. Zippette's pregnancy has put a lot of pressure on Zippy to shape up and get organized. As she has become less mobile, he has had to take on more of the household chores, like vacuuming the rugs and dusting the server clusters. However, this puts him in closer proximity to the servers' keyboards than is absolutely safe, like within arm's reach. Much like a cat that won't stop pawing at a bumblebee, Zippy just can't resist messing about with computers.
An auto mechanic I knew once told me: "You only have to be 10 percent smarter than the thing you are working on." Unfortunately, Zippette's Linux super-cluster can wipe the floor with chess grandmasters and Zippy is nowhere near that high on the food chain. Maybe if he had read a manual once in a while he would have fewer disasters. Then again, most manuals assume that you do not have to tell the user to remove Disk A before inserting Disk B. Zippy killed a perfectly good slot-loading CD drive that way last year. Since then he has felt that manuals are not as complete as they should be and just stopped using them entirely.
Linux, however, is not something you should trifle with in ignorance, particularly when it's your wife's current pride-and-joy and she's enduring mood swings that make pyroclastic clouds seem quite safe by comparison. Undaunted by the prospect of great pain if things went awry, our hero decided to see if he could lift her spirits by upgrading her 22-inch flat-panel LCD monitor to a 43-inch plasma screen. It seemed simple enough: just download the appropriate drivers, find a way to load them, and swap the cables. Five hours later, however, I got the inevitable call for help.
Now I know a bit about Windows and a bit more about Macintoshes. I have also used Unix systems a little. But until Zippy called I've never actually tried Linux myself. He has a lot of faith, however, in my ability to debug even unfamiliar systems. He also hoped that since I have a black belt in aikido (Japanese art of nonviolent self-defense) that I would be able to defend him from his wife if things did not work out.
Here is what had happened. The two driver files he needed downloaded easily and he installed them using a program called Kpackage. (Kpackage is a Kopernicus Desktop Environment (KDE) tool for installing, viewing and uninstalling packages.) Many messages flashed by too fast for him see what was being written on the screen. Emboldened by his apparent success, he then tried to update something called the "X11 configuration." A few clicks later he found something called KickStart. Click. Oops. No more video.
Ctrl-Alt-Backspace. Nothing. Ctrl-Alt-Backspace. Nothing. OK, that didn't work. Then he tried Ctrl-Alt-Del. Text appeared telling him the system was "going down." When it rebooted the monitor continued to flash. Obviously he did something wrong with X11, which is part of the X-Windows graphic interface sub-system for Unix variants. But what? Only then did he try the manual, which gave him the correct procedure for reconfiguring X11. Unfortunately, the manual assumes you followed its procedure, so of course there are no helpful instructions on how to recover from the certain disaster that results when you don't.
So he called me. Unlike Zippy, I know when I am in over my head. I gently pried him away from the keyboard and we logged into a Linux Web site using another system. Thanks to some help from a person with the chat handle "TuxGod2" we had the cluster's video working again in less than 15 minutes. Zippy's little misadventure highlights both sides of the open source equation: it may be cheap and easily available, but it is not always easy.
Protecting the Source
Okay, so we have "free" software. Or do we? Software companies use copyright law to protect the rights of software owners and developers. Licenses control the exercise and transfer of those rights to other parties. Under copyright, software is protected when it is first created and fixed on media, regardless of whether the creator or author explicitly reserves the copyright with a notice or registers it with the U.S. Copyright Office.
Copyright law secures and protects the right of software vendors to copy, distribute and prepare derivative works. These rights are separable and can be transferred to other parties by license. Proprietary software vendors have a financial interest in maximizing their return on investment by prohibiting modifications to products, limiting the number of copies a customer can make and restricting their distribution. This protective attitude is the basis for why most software companies want customers to pay for every copy of the software they install. They are just now getting around to hard-coding this though the use of Internet-based product activation codes, Windows XP is a recent example.
Open source software, like proprietary software, is protected by copyright and can be restricted by license. However, open-source vendors are usually much less restrictive and allow users to run and copy the software, modify it and distribute the modifications under certain conditions that vary depending on the license. The degree of freedom given to open source users depends on who is distributing it. Developers or vendors that wish to distribute software widely relax license restrictions and let users copy, modify and even distribute the modifications with minimal restriction. For example, when AT&T stopped providing Unix source code free to some academic institutions, programmers at University of California, Berkeley, developed the BSD (Berkeley Software Distribution) operating system.
They wanted to distribute their Unix clone as widely as possible, so they released its source code with a license that lets users modify it and distribute the modified version with or without source code. The permissive BSD license worked. The open-source community modified the code and returned it as FreeBSD, OpenBSD and NetBSD. Sun Microsystems originally distributed BSD free with their hardware and later turned their version into a proprietary OS. BSD code is the core of the new Apple MacOS X and can also apparently be found in Windows NT.
Penguins Rule
Despite BSD being there first, the Linux operating system is probably the first thing that comes to mind when people think about open source. Linux is a free variant of Unix created in the early 1990s by Linus Torvalds, who was a college student in Finland at the time. While Linux is not the only free Unix variant available, it is arguably the most famous at this point in time. What distinguishes Linux from all the other free variants is two things. First, several big players, most notably IBM, Hewlett-Packard (HP), and Compaq, are backing it. IBM announced they would invest $1 billion in Linux as a replacement for their own, admittedly aging, mainframe operating system. Early this year, IBM claimed they've already made back their investment.
Where did they make that kind of money? Well, their list of Linux customers includes clothing retailer, L.L. Bean; digital animation studio, Pixar; department store chain, Boscov's and financial services giant, Salomon Smith Barney. These are not bit players. L.L. Bean is using Linux to manage their customer e-mail notification system, which now can send 30 e-mails per second to customers around the world. That's six times faster than their old system. They have plans to move more of their operations to Linux later this year. How did L.L. Bean go about becoming a poster child for open source? They started by buying a copy of SuSE Linux for $60 at BestBuy. Now that they have it tuned to their liking, they can copy their customized version to as many systems as they want without having to worry about licensing fees taking a bite out of their budget.
Circle the Wagons!
The second thing that distinguishes Linux from its peers is the reaction it is generating from Microsoft, which has apparently decided Linux is a threat that requires a full-scale training program for their sales force on how to combat it. I will not get into any of the Windows vs. Linux debate here. That is better left for the faithful on both sides to fight out in the publications of their choice. But the simple fact that the most powerful software company in the world has seen fit to target Linux (and doesn't seem to be having much luck in slowing it down) is significant.
But is Linux really ready for prime time? According to reports from the LinuxWorld Expo earlier this year, HP's Chief Executive Officer (CEO), Carly Fiorina gave a keynote speech where she talked about 2002 being the "breakout year" for Linux. She shared HP's plan for the "mainstreaming of Linux" and how she sees her company's role in helping to give credibility to open source software. Fiorina's bottom line was, "We want to help our customers take full advantage of the open standards of Linux."
The use of Linux is growing rapidly, particularly outside the U.S. The most publicly recognized use of Linux is as a commercial server operating system handling "back room" computer tasks like sending and receiving e-mail, hosting Web pages and downloadable files, funneling "print" commands from desktops to networked printers, hosting databases, and all the other mundane things networks do that are hidden from public view, but are absolutely necessary for efficient networking. Walk into an office in any corporation today and, even if you see Windows on every desktop computer, there is a good chance (one that is increasing every year) that part (or all) of the underlying infrastructure that connects all those desktops is running Linux.
One area where open source software may soon dominate is Web hosting. According to the January 2002 Netcraft survey, the open source Web server software Apache runs just over 63 percent of all Web operations around the world. Compare this to Microsoft Internet Information Server's (IIS) 26 percent share. That's not too shabby for a product whose name was derived from "a patchy server," which refers to Apache's early days as a hacker project with new software patches released almost daily. Apache 2.0 was released earlier this year and some in the industry are suggesting it may be the final death knell for IIS.
As Apache 2.0 is fully Windows compatible, Windows users now have a Web server that is specifically designed to support the XP platform and is arguably more reliable and secure than IIS. Ironically, however, the biggest beneficiary could be Microsoft. Until last fall, the company had been trying to get its WindowsNT users to upgrade to Windows 2000 with mixed success - in part because of Web management issues. Their current desire is to get users to upgrade to Windows XP. In an ironic twist, Microsoft now stands a much better chance of selling Windows XP to people who prefer using Apache to manage their Web sites.
The only potential downside for Microsoft might be the demise of IIS. However, this would not have much financial impact because the IIS Web server is normally bundled "free" as part of Windows. In addition, Microsoft may be quietly acknowledging defeat: IIS is not available in the Home edition of Windows XP and the Professional edition, by default, installs without it. Between that tacit withdrawal and IIS's highly publicized security problems, it's possible that IIS may eventually wither away under the Apache assault. Combine all the preceding with the fact that Oracle, one of the worlds leading database management products, runs on Linux, and you have what looks like an open source volcano about to erupt on the business computing scene. However, there are some fairly substantial problems that the open source movement must surmount before that can happen.
Sticky Is As Sticky Does
One area that is still an Achilles Heel for open source is the lack of office automation and publishing software for the desktop. For anything that requires printing your work on paper, there is really nothing for Linux that rivals the publishing software available on other platforms. There is, for example, no Linux version or equivalent substitute for staple publishing programs like Quark Xpress or Adobe Photoshop. While there is a potential replacement candidate for Microsoft Office in Sun's free StarOffice (a.k.a. OpenOffice) suite, the individual applications in the free version don't quite measure up yet to what people are used to in MS Office and it does not include an equivalent to MS Access. I've used StarOffice, and I like it. I even sneaked a few papers and slide shows created in StarOffice into official briefings, just to see if anyone would notice. No one ever caught me. But every once in a while I actually needed some of the features in Office.
Another problem is that Linux is not really designed for desktop use by the average user. Most people can navigate Windows without too much hassle. Linux does have some graphic interface shells, notably Gnome and KDE (a network transparent contemporary desktop environment for UNIX workstations.), but they do not insulate the average non-technical user from the system as well as Windows does.
I will also state a personal mea culpa here: I am not ready to migrate my home computers to Linux. I have a copy of Linux that I have used to peek under the hood. However, my home computers are for the family to work with simple things like e-mail, letters, and digital photos/video. The OS that came with the systems is good enough for us, so it's unlikely I will be complicating my life at home with Linux any time soon.
Personal comfort aside, the biggest hurdle to installing Linux in the enterprise is the stickiness of proprietary products in general, which tend to herd users towards proprietary file formats, scripting languages, application programming interfaces and executables. Microsoft software, for example, comes with default proprietary formats for its various Office applications. You can save a Word file in rich text format (RTF), but why would anyone bother? Add in Visual Basic scripts, MAPI, and ActiveX, and you have a fair amount of built-in inertia for the Windows platform. In addition, Microsoft products are, for the less technically proficient, easier to install and manage as prepackaged solutions than most Unix applications, which is very attractive to people who think they have better things to do with their time than manually upgrade their graphics sub-systems.
However, despite all this, I think in the next few years we'll see more and more operations going to Linux, and one of the contributing factors will be Microsoft's own plans. Based on what I get paid versus the CEO of HP, I may not be in the best position to disagree with Ms. Fiorina on when this will happen. But I think the real revolution will come after 2003, not this year. Here's why: 2003 is the year Microsoft is apparently planning to stop full support for Windows 98 and NT, and slow down patch and bug fix activity for all versions of Windows older than XP. It's a pattern they've followed in the past that encourages customers to upgrade to supported versions.
Because of this, 2003 will be the year that a large number of Windows users are going to be faced with the choice of either buying new hardware that will run XP or moving away from Windows entirely. I believe at least some of these people (and organizations) will choose Linux or BSD as an alternative for some (or even all) of what they do.
Cost Comparison
Okay, let's get down to the numbers. What does it really cost to switch from Windows to Linux, and is it really worth it? There was a commercial running on network TV recently that went something like this: As lights flash and sirens blare, a frantic manager leads some police officers into a vast, empty room yelling: "The servers, they stole all our servers!" Out of the emptiness emerges a slightly rumpled techie who gestures toward a single refrigerator-sized box in the far corner: "No, we moved everything onto that one. Gonna save us a bundle. I sent out an e-mail." The words "Servers running Linux" flash across the screen.
But, as we should ask of all commercials, is it true? Transferring everything to one z-Series mainframe isn't something you do with just a phone call and a credit card. There are still licensing, maintenance, training and other sustainment costs to consider. Did that techie really save a bundle? ConsultingTimes.com (CT) checked the commercial's vitals with Jimmy Lee, Director of Professional Services for Equant, a global network service provider that runs the world's largest IP data communication network. According to CT, Equant was among the first organizations to implement Linux on the S/390 and z-Series mainframes and migrate its customers to open standards, so Mr. Lee is one person who can reliably fill the IBM ad with real numbers. Here is the bottom line of Lee's comparisons of an Intel/Exchange system versus an IBM/Linux system for 5,000 mailboxes:
- The Exchange/Intel solution costs $ 8.65 per user per month.
- The Linux groupware solution costs $ 2.02 per user per month.
There is one caveat: the Linux solution assumes you already have a mainframe in place. Add and amortize the capital investment of a mainframe to adjust those numbers to match your situation, as appropriate. However, if the possibility of saving up to 75 percent of your total cost of ownership appeals to you, please go to http://consultingtimes.com/Serverheist.html and read the entire cost study. It's worth a look.
Conclusion
Open source software is not for everyone. It's often like owning your own custom-built hot rod: as long as you know how to tinker with what's under the hood, it works just fine. Migrating in any significant way to open source requires a great leap of faith. In the interest of helping everyone make informed decisions, I would like to share some of my favorite sources of information.
I have read many accounts of various open source migration efforts. The best example I've found so far was published in the Jan. 1, 2002 issue of CIO magazine. CIO is not, by any stretch of the imagination, a "hacker-zine." It is about as Main Street businesslike as you get in the IT world. The article, "How to Run a Microsoft-Free Shop," by Scott Berinato, outlines a 12-step approach for weaning your organization from dependence on Microsoft products.
Another article worth reading is "Escape from Redmond," published by Jerry Pournelle in the Jan. 7, 2002 issue of Byte.com. Pournelle is an absolutely meticulous IT critic and one of the most reliable sources I've read for reviews of computing technologies over the last 10 years. Pournelle is bluntly honest about what he thinks Linux is (and isn't) ready for.
Finally, no tour of the open source world would be complete without visiting Newsforge, a Web site that links to open source news articles and opinions from a variety of sources. Newsforge has been on my daily must-read list since I became interested in open source about four years ago. If you're thinking about trying open source, I recommend you spend some time there.
I don't seriously expect everyone who reads this to go out and "rip and replace" their enterprise systems with Linux, Apache, Sendmail, and StarOffice. If you're happy with your current systems and what it costs to run them, carry on. I do expect you to think about it, though. In a world where computer technology burns through a new generation of products every 18 months or so, you never know when you might need a new alternative.
Happy Networking!
Long is a retired Air Force communications officer who has written for CHIPS since 1993. He holds a Master of Science degree in Information Resource Management from the Air Force Institute of Technology. He is currently serving as the Telecommunications Manager for the Eastern Region of the U.S. Immigration & Naturalization Service.
The views expressed here are solely those of the author, and do not necessarily reflect those of the Department of the Navy, Department of Defense or the United States government.