OPINION: Linux Is Not Ready For the Enterprise

A few months ago I was asked by a major publication to take the Windows side of a Linux vs. Windows debate. The arguments raised by the writer on the Linux side revolved around fuzzy political, almost religious, concepts like freedom of choice and the need to be counted in the battle against the Microsoft monopoly. I found this somewhat entertaining since she is a CIO and IT organizations generally don't give their users a lot of choice.

In my response, I systematically pointed out that the justification was based on something that sounded more like a religious belief than business benefit, and such a justification was a short path to unemployment if it was the basis for a highly critical project and the project failed. This would be true regardless of what platform we were discussing.

The writer on the Linux side objected, not to the facts, but because my article would have put her job at risk. Because of the CIO's job concerns, the debate never ran; this was unfortunate because through debate we gain better understanding of the risks involved in any decision. And we can satisfy the need to develop a solid business case for a decision in case the competency of the decision maker comes into question due to an unforeseen problem later in the process.

As an analyst I have to be able to argue both sides of a position because often we are asked to step in and help justify decisions that have already been made (not a recommended practice but one too often used). The Linux folks can certainly argue, sometimes viciously, for their platform. But when asked to come up with strong, well founded, business arguments, they too often go mute, or begin what often turns into the mother of all flame battles with little real content. This makes it very difficult to have an intelligent discussion, and discussion is critical to a well-considered decision. What follows is clearly not unbiased, it was originally written specifically to be an argument against Linux and, just as clearly, there is a pro side that I will leave for others or a future time. My personal goal is not to get people to stop using Linux but to really think though the decision, and if they decide to deploy the platform, be better prepared to defend it and protect themselves and their company.

Linux is a good platform, well understood, and widely distributed. It is backed by practitioners, particularly students, who defend it with great vigor. It is actively replacing Unix and MacOS as the contrarians' platform of choice and owes its success largely to the failures of these earlier platforms to run on multiple hardware vendors' systems in a consistent fashion. Linux is also attractive because it is not from Microsoft.However, Linux is in many ways a throwback to more primitive systems. Not only is it repeating the mistakes of its predecessors, it apparently is introducing a brand new set of problems, having to do with intellectual property.

Linux returns us to the days when the computer industry was a cottage industry, when programmers and engineers drove the industry and users were overwhelmed by the difficulties in getting what they wanted done. Linux is anything but a high-volume platform and applications are generally highly customized and more closely tied to the people who developed them than to the users that will live with the resulting software. This is not so different from the mainframe world we, for the most part, left behind us by the late 80s. Linux software is perceived as "free," which helps make it a strong counterculture platform. Most of the value, and cost, is on the hardware and services that surround it. This idea of attaching a low value to software is a throwback to the beginnings of computing, and only changed after a protracted fight between IBM and the U.S. government over unbundled hardware and software.

You can see why programmers love Linux: It gives them apparent power, control, freedom, and flexibility at a low initial cost. But this is a zero-sum game and when one group gains power another often loses it. Granting this kind power to any one group, particularly a staff function, is often not in the best interest of an enterprise, which needs to focus more on business needs, including time to market, and competitive advantage. Unless the business is a software business, putting much of the cost, talent, and control into the hands of software developers is generally inconsistent with business needs and company goals. Even in the case of a large software company, putting top human resources into the development of software for internal use, rather than developing a marketable product, would seem counter to running a successful business. We are in a period where specialization, not generalization, appears to separate the successful companies from those whose longevity is not so certain.

If you are in manufacturing, you likely have a policy to buy off-the-shelf manufacturing equipment whenever possible, because it is vastly less expensive than having custom machines installed. If you are in field services, you typically don't have custom trucks built, but buy from a large manufacturer and then make relatively minor changes to fit your specific needs. If you are a successful builder you ask your architect to minimize the custom components used so that you can maximize your profits. During the rapid growth of Linux during the Internet years, the idea of containing costs and minimizing risk was something that many seemed to forget. With the failure of the dot-coms the concept of solid business plans and solid cost containment practices are once again more like laws than suggestions, changing dramatically the way decisions are viewed and approved.

While I've seen a lot of discussion of Linux licensing, the general impression seems to be that most of it has little relevance to an IT organization in a company that is not, itself, in the software business. Recently SCO (formerly Caldera) began to change that perception and began to demonstrate what could easily become a nightmare for the Linux community. Whatever the outcome, it is clear that not enough consideration was put into the protection of intellectual property in the creation of this platform and that, currently, much of the risk seems to be getting passed downstream to enterprise users of the product, rather than upstream to the distributors. SCO earlier this year issued a letter to 1,500 of the world's largest corporations, warning that Linux users might be liable for intellectual property violations in Linux. The warning is contrary to the normal course of things, where vendors protect their customers from intellectual property liability. (Users of Linux from IBM and Hewlett-Packard are more likely to be protected from SCO's lawyers; IBM is already the subject of a lawsuit by SCO, and the nature of the HP license provides HP's users with protection.)IT organizations are simply not equipped to deal with intellectual property theft issues within a product they have deployed widely. They have neither the legal expertise nor the budget to even properly assess the risk, let alone effectively mitigate it.

The key licenses that surround Linux, for the most part, have yet to be fully tested in court. These initial tests seldom go the way the drafters intended, because judges are not technical and the issues are complex. With the proliferation of firms whose existence is supported solely by the protection of intellectual property they have acquired, these tests are fast approaching and will likely, over time, identify the weaknesses in the current approach. The SCO-IBM lawsuit is the latest example of these kinds of tests, as are NTP's patent lawsuit against Research in Motion, Intergraph's litigation against Intel, and even Sun's Java lawsuit against Microsoft. The market may soon be defined by the ability to litigate rather than the ability to develop, and products like Linux, which have a weak defense, may simply not survive this market phase.

One of the things that most concerns me, because it was major failing in previous anti-establishment (read "anti-Microsoft") initiatives, is the behavior of the most visible advocates for these alternative platforms. Microsoft has clearly been blessed with challengers who apparently never learned not to run around blindfolded with sharp objects pointed at their own hearts.

Many Linux users are outspoken and militant. Like their OS/2, MacOS, and Unix predecessors and counterparts, they make personal attacks and broad public statements. Some avid Linux defenders make statements that are unprofessional, and filled with words we wouldn't accept in the workplace. These defenders are clearly doing damage to the credibility of their effort. Moreover, they bravely, or foolishly, often identify themselves by company and position. Public "debates" about Linux contain behavior that could easily violate HR rules as the arguments drift into language that has become unacceptable in the enterprise and has little to do with the topic being debated. These arguments create embarrassing situations which could, in some cases, percolate up to board levels.

Not all Linux folks are like this, any more than their predecessors were. The bad behavior is limited to only an increasingly vocal, and apparently growing, minority. But an enterprise, by nature, has a huge number of employees who are held to solid policies about appropriate workplace behavior that have to be enforced. Any product that promotes behavior that violates some of the most critical of these policies should be on the short-list of things to be avoided in an enterprise. Clearly any "alternative" platform that has backers who can't control their language, or worse, use methods which now are classified by several governments as terrorist acts, should be on the list of things you would like your competitors to use but would avoid yourself like the plague.In short, while Linux is technically a very competent product, it still lacks the necessary maturity for a mission critical enterprise deployment. It does have a place as solution for small companies who, themselves, occupy cottage industries and where a handshake is all the contract you really ever need; in an enterprise, unfortunately, cost controls and solid policies that put the business first must take precedence and place Linux off the list of consideration, possibly forever, for many enterprises.

OK folks, take a breath, let it out, and now take another. Remember this is a "think" piece and not an offer to write me with your sincere opinion that I am Bill Gates's love slave (it's already been done anyway), I know Bill, and trust me, we aren't that close. Assuming someone doesn't put a bomb under my car I'll be back with another column in a few weeks. The working title for that one is "What's Wrong With Microsoft."

Rob Enderle is a Research Fellow at Forrester Research and his research activities have covered the IT space from vendors like IBM, Dell and Compaq, to technology companies like Intel and Microsoft. Related projects have involved positioning products for the IT market, staffing, training, resolving quality control problems. vendor selection and risk avoidance. Other ongoing projects include studies on vendor satisfaction, market predictions and personal computing product selection.

Rob joined Giga (which was recently acquired by Forrester) in 1995, bringing with him more than 25 years of experience in the high tech industry. Prior to that, he was with the Client Server Software service in Dataquest, where he was responsible for operating system forecasts and related technology trends. He came to Dataquest by way of IBM/ROLM and still longs for the day when he can once again work for a firm that has a "Great Place to Work" department.

Rob spends his free time building custom personal computers, hot rodding his wife's car, and trying to figure out his house, which is fully automated and was recently converted to solar power.In 2001, Technology Marketing named Rob as the most influential industry analyst.