Four years ago, without really realizing what it was doing, Minolta Canada Inc., a subsidiary of the Japanese camera and business equipment manufacturer, took a bold step into the vanguard of the Open Source movement. Even today, Minolta remains among a relative handful of mainstream enterprises pioneering the use of Open Source solutions in areas other than Internet and e-commerce.
Open Source is the term a wired community of software developers chose to describe the platforms and products they were using and creating for which the source code was made freely available, usually over the Internet. This is in stark contrast to commercial platforms where the source code is a closely guarded trade secret.
The Open Source movement was driven initially by a virtual mob of far-flung developer zealots. Even the most famous, Linux inventor Linus Torvalds, considers himself a first among equals. More recently, hot young companies such as Web server vendor Apache Digital Corp. and Linux distributors like Red Hat Inc. and TurboLinux Inc. have emerged, bringing commercial viability. The movement has gained surprising momentum, mainly fuelled by the boom in e-commerce and other Internet development projects.
But it was in late 1995, before that boom had even begun, that Minolta systems manager Anthony Pinto gave his okay to a nothing-to-lose project proposed by a new supplier, Brampton, Ontario-based Unix software developer Starnix Ltd. Starnix offered to take a cast-off 486SX PC and, using Linux, the then little known open source operating system derived from Unix, turn it into a working e-mail server on Minolta’s Novell network. It would serve the company’s small office of 30 or so employees.
“It was an experiment,” Pinto says. “They took a box that was just sitting there, redundant. There was virtually no cost involved. So we said, ‘Sure, go ahead.’ When they completed it, it could do much more than what was expected. Soon after, somebody said, ‘Let’s put in a proxy server,’ and it just grew from there.”
Today, Minolta has three Linux servers on its Novell network, including one that handles EDI transactions and a Web server for the company’s slick Web site. A lone Unix server runs the company’s legacy ERP system. After the initial e-mail experiment, Pinto’s camera division added employees from Minolta’s much larger business equipment division with which it shares office space. The server now supports over 300 users – a few too many for the original 486. More Linux-based projects are in the works, Pinto says, but he can’t talk about them yet.
Open Source solutions have so far had tremendous success among Internet service providers and e-commerce-centric organizations. A survey in September by Netcraft, a UK-based Internet consulting firm, found that of the large number of Web sites responding to its poll, over 60 per cent were using servers running the Apache server operating system. Industry research firm International Data Corp. anticipates Linux shipment growth will outpace all other server operating environments through 2004, with shipments increasing at a compound annual rate of more than 28 percent, from 1.3 million in 1999 to 4.7 million in 2004.
Little Enterprise Headway
But Open Source has so far made little headway in the enterprise and less in applications not related to the Internet. Minolta with its Linux EDI system is still an exception. But that could be changing, and advocates such as Starnix president Evan Leibovitch and consultant Richard Reiner, CEO of e-business service provider FSC Internet Corp. of Toronto, say it definitely should change. They cite some compelling reasons.
One of the most obvious, says Leibovitch is that because Open Source platforms are open, “it means a user can’t be held hostage by a supplier”. They don’t have to upgrade when the developer says they must. They don’t have to wait for the developer to make changes to a product. With source code in hand, a customer can make changes himself – or get a Linux software house to do it.
And they’re not locked into a single vendor. There are lots of other Linux vendors out there if a customer is unhappy with service or support. Leibovitch points out that the perception of there being many incompatible ‘distributions’ of Linux is just that, perception. Linux versions from RedHat and Caldera, for example, the distributor Starnix uses, are “95-percent compatible,” he notes. “They differ mostly in implementation of the graphics interface. But if you like one vendor’s graphical interface solution, it’s a trivial matter to implement it on another’s system.”
Despite a widespread perception of Linux and other Open Source platforms and products as software mavericks, and the movement itself as vaguely anti-establishment – even Marxist – advocates like Reiner insist that rock-solid stability and reliability are among the hallmarks of Open Source solutions. “It’s typically exceptionally good quality software,” he says. “And it’s usually exceptionally stable – more so than with comparable commercial packages.”
The comparison is most often drawn beteween Linux, Apache, et al on the one hand, and Microsoft NT or 2000 on the other. Reiner prefaces his remarks by saying, “I don’t want to get into Microsoft bashing.” But in fact it’s hard for Open Source enthusiasts to resist it. Microsoft is the bete noir of the movement, perceived by some as the very model of the monolithic, user-unfriendly software-development regime.
“A number of organizations that have gone the Microsoft route have had stability issues,” Reiner says, repeating what any Open Sourcer will tell you – and lots of Microsoft users would confirm. “But on a Linux platform – except in the case of hardware failure – it’s absolutely mundane to see systems up and running for 500 days.”
One of Reiner’s clients, John Bacopulos, chairman of Toronto-based Web Offset Publications Ltd. and LinkPath Internet Network Corp., can vouch for the reliability of the Apache servers FSC implemented for him a year ago. Bacopulos went to FSC when he was developing LinkPath, a Web site and service that helps print-magazine publishers and adverisers connect with their readers via the Web. FSC persuaded him to use Open Source solutions – Apache servers and the PHP3 scripting language.
“We just wanted something that was rock solid,” says Bacopulos, who admits he did not have a deep understanding of the technology issues involved, but trusted FSC. “That’s what we got. We’ve had aboslutely no crashes.”
Minolta’s Pinto tells a similar story. “Up-time is almost 100 per cent,” he says of his company’s Linux-based systems. “Come to think of it, I’ve never seen any major or even minor down-times. Obviously you would if the connection drops – but we’ve never been down because of the software.”
Why are Open Source platforms so stable and reliable? Because of the very nature of how the software is developed, Reiner says. With commercial products, a relatively small number of programmers work on and review code. But because source code for Open Source platforms and applications is available to anyone who wants to download it, they have the advantage of tens of thousands of developers all over the world working in concert – many of them vitally interested users.
These volunteer developers contribute ideas and chunks of code. But most importantly, they review code, root out bugs, and suggest fixes. And just by the sheer number of people involved, they can do it very rapidly. This is particularly important when it comes to security issues.
Indeed, one reason Open Source solutions have found favour in the Internet space is because users can be more confident about security, Reiner says. Fewer security glitches are found in Open Source products, and those found are fixed faster, he claims. Reiner has a special interest in this area. The FSC division with which he is closely involved is SecureXpert Labs, a firm that advises Internet companies and mainstream enterprises such as Shoppers Drug Mart, BCE Inc. and Sun Life Financial on security matters.
Some commercial platform and application developers do have an excellent track record when it comes to responding to and dealing with security breaches in their products, Reiner says. But because they can’t afford to have customers and prospects lose confidence in their products, too much of their response to a discovered security breach is managing the discoverer and the information about the breach. Reiner knows – his company has found a few security holes in commercial products over the years. Commercial software developers also have limited resources to throw at solving the problem.
“The typical time to fix [a security problem in some commercial products] is best measured in months,” he says. “But when you’ve got tens of thousands of developers reading the source code every day – for their own enlightenment or to scan for security problems – it’s done a lot quicker. With Open Source platforms we’re getting very close to real-time security fixes.”
Reliability and stability are not all Open Source platforms and applications have to recommend them. Cost of ownership is also lower. Leibovitch says that in competitive bidding situations, his company consistently beats out quotes from other systems integrators that use commercial platforms – both on cost and performance.
For starters, there is the cost of licenses. If you download Linux and other Open Source products, even from commercial sites like Red Hat’s, they’re free. This is in contrast to the thousands of dollars in license fees companies pay for commercial products. Vendors like Red Hat and TurboLinux do sell Linux on CD, but they’re really selling the extensions and applications bundled with it – and the convenience of not having to download. The price is often less than $100, and it’s a one-time charge. Customers are free to duplicate the code as many times as they want.
Cost was certainly a factor for Pinto. He didn’t even have to do a business case on the first e-mail server project. “There was really nothing to analyze,” he says. “If we’d bought NT, it would have blown the budget. It’s a pretty straightforward decision, especially for a small office. But we’re not that small anymore. [However, the technology] is scalable.”
Comparisons are inevitably drawn with NT. There is a hidden cost to using NT, Reiner believes. “NT shops have learned from experience that it runs better when you do not run more than one application on a single server,” he says. “That’s an issue that just doesn’t exist in the Linux world. If the hardware capacity is sufficient, there’s no reason not to run more than one application per server.” Extra servers cost more money.
Cost, though, remains “a secondary motivation,” Reiner says. Most companies are more interested in the quality, reliability and security of Open Source solutions. Of course, as Pinto points out, highly reliable and bullet-proof systems cost less to run and maintain as well.
Not For Everyone
So if Open Source solutions are such hot stuff, why are corporate IT departments staying away in droves?
Even advocates like Reiner and Leibovitch admit they would not – do not – recommend Linux everywhere. Open Source solutions are not strong on the desktop, for example. Reiner admits he doesn’t recommend Linux for the desktop except where users are technically sophisticated. As Minolta’s Pinto says, “Microsoft still rules on the desktop.” But this is changing. The Linux-based Gnome graphical user interface is evolving rapidly and has the potential to be a far more sophisticated platform than Windows, Reiner believes. Eventually.
He also would not recommend Open Source solutions for high-end, large-scale applications. Linux is still behind in areas such as high-availability systems and CPU clustering, for example.
“If you want to do something as simple as shared storage clustering, you’re not going to find Open Source solutions,” Reiner says. “Except very green ones. And no CIO or systems manager is going to think of using version 0.95 of anything. That kind of enterprise functionality that is taken for granted on the Sun [Solaris] platform is just barely available in Open Source.”
This too is changing. TurboLinux in particular has products that address the needs of large enterprises. And all the major relational database management systems, including Oracle, are now available for the Linux platform. Leibovitch points out that most of Oracle Parallel Server is now certified for Linux clustering. “And that is something of a litmus test for any platform that aims to be any good for enterprise use,” he notes.
It may take awhile for this message to get through, though. In explaining why large enterprises are typically not bringing Linux in through the front door, Peter Poolsaar, leader of the infrastructure and network integration practice at Cap Gemini Ernst & Young Canada Inc., says that “ultimately the choice of platform is dependent on the application need. And a lot of the big databases don’t run on Linux.”
Poolsaar also points to the fact that many enterprises have policies that preclude buying from any but established vendors. Yet IBM and Silicon Graphics Inc. (SGI) are both very active in the Open Source movement. That message too, has apparently not sunk in yet. IBM has committed to embracing Linux across its entire product line – it even runs on high-end S390 mainframe servers. Both companies have introduced journaling file systems for Linux and their own clustering technologies. SGI’s ProPack 1.2 for Linux is a system overlay that adds many large-system capabilities to Linux on SGI hardware.
CIOs may just not be comfortable yet with the idea of Open Source platforms, Poolsaar suggests. There is the perception that support is a problem, that there is no reliable corporate entity accountable for support – notwithstanding the Red Hats, Calderas and TurboLinuxes of the world that are very much in that business. This diffusion of accountability in the Open Source movement, which is fundamental to the concept, is difficult for many IT executives to grasp.
In Need Of Enterprise Implementations
It would help if there were some prominent enterprise implementations. When they’re considering a solution, corporate decision makers typically want to see installations similar to their own. “If the vendor can’t point to a number of large implementations, the risk profile for that solution goes way up,” Poolsaar explains.” The cost benefit picture changes. You start asking, what will it cost to take that risk?”
Even among mid-size companies, good examples of non-e-commerce-related Open Source projects are few and far between. Reiner mentions a real estate company that built its building-information system on a Linux platform, and a waste management company for which FSC is building a new dispatch system. But neither company would talk on the record about what they were doing.
Besides the “intense” activity in development areas where Open Source currently lags – high-end applications and desktop in particular – there is one other trend that will push Open Source into the enterprise, Reiner believes. That is the move towards browser-based applications – applications that use a standard Web browser as the main user interface. The system his company is building for the waste management firm is one example. If you’re using an Internet-based front end, there is some logic to going with an Internet back end as well. “And then all these tools that have been the forte of Open Source become applicable,” Reiner notes.
Will Open Source move into the enterprise in a big way? We’re guessing it will take a bit of time for it to happen. Even Pinto, for example, won’t consider moving to Linux for the company’s main ERP system – not yet anyway. The existing system works fine. But as Poolsaar notes, Linux is leaking in through the back door – it’s a favourite among developers because of the relative ease of development and reliability – and through e-commerce projects, which are endemic.
And if Open Source solutions are as good as advocates like Reiner and Leibovitch say, it seems almost inevitable they will gain converts in the enterprise.
Tony Martell is a freelance writer specializing in information technology and IT management. He is based in London, Ont.