Desktop Linux: The Final Hurdles

I remember the major turning point that took Linux from an OS that Microsoft used to call competition during their anti-trust case to a serious commercial operating system. At the time, Linux users numbered two million and Microsoft users numbered 300 million. I used to say Linux makes more news than anything else and I owned a Linux company.

In April 1999, D.H. Brown Associates, Inc. published a report called Linux: How Good Is It? Hardly any archives exist today mentioning that story with the possible exception of this C/Net News article.

The study dinged Linux for lacking features needed to make it a serious consideration as an operating system. The report said that Linux was good for file and print servers, Web servers, some scientific computing, and thin client computers. But, the DH Brown report said Linux lacked support for computers with multiple processors; failover and a "journaling" file system needed to reboot a crashed machine without having to reconstruct the system files.

I attended a briefing on the report by Linux International and learned that the kernel developers started addressing the issues before the report's official release. As we know now, the open source community responded rapidly and the Linux server has taken a commanding share of the server market. The DH Brown report became the Linux server roadmap placing it among elite operating systems.

What About the Desktop?

The Linux Desktop sits at the foothills of success. I have walked in the trenches in numerous market sectors and would like to report what Linux needs to become a serious competitor to Microsoft. Like D.H. Brown, I don't intend to pull any punches. Linux is a very good desktop but it cannot play seriously in the market until Linux companies take customer feedback seriously.

Large and small organizations want a Linux desktop. Many face the expiration of their Microsoft Software Assurance contracts and do not want to renew them. The value proposition of Linux is compelling. The platform requires less administration, fewer support personnel and offers unmatched security.

You cannot fault corporate customers for failing to try the Linux desktop. If you could see the lists of pilots of which I am aware, you might fall out of your chair. If Bill Gates saw this list, he might consider offering a Linux desktop. But, the companies trying Linux do not want to alert Mr. Gates for fear of reprisals –audits by the Business Software Alliance.

Two complaints seem universal: Some Linux Desktops come ready to use out of the box but fail in their due diligence. The others pass due diligence but do not come ready to use out of the box.

Corporate Requirements

In the fall of 2001, I received a call from an IBM reseller working with one of largest investment banks on Wall Street. During the 9/11 tragedy, the company lost its email capability and could not restore its mail on its Microsoft Exchange servers. As a consequence, the company faced stiff penalties from its regulators.

IBM had made my firm a Business Partner for porting a replacement for Microsoft Exchange to Linux on the S/390 mainframe. The product worked and generated IBM's first domestic Linux reference account, Winnebago Industries.

The IBM reseller handed me a list of requirements that included a deeper level of interoperability for shared calendars. Other than that, the customer had tried our server, liked it and only wanted a little more functionality. We made changes to the product and met the requirement. But it didn't end there.

If you follow the sequence of events following our meeting the requirements, they can reveal the nature of meeting widespread corporate needs. Between the IBM reseller and my Linux firm we satisfied the requirements the investment bank: The ability to restore mail during a catastrophic event and their need for robust meeting management. According to our contract and the criteria for success, we should have won the business and started deploying the solution. The customer, however, wanted more. They wanted a transparent solution so their administrators wouldn't know the difference.

We followed change after change in requirements for approximately nine months and the customer refused to pay us anything. They said they were helping us make a better product. Additionally, the customer demanded that we meet the changes or lose the business. Try servicing a zillion dollar investment bank with all your developers while you attempt to live off revenue.

As a member of the board of directors, the COO and director of development, I put my foot down after we met the first requirements. I said, drop the account and service our existing base. I was out voted and eventually left nine months later when the customer finally said that the product wasn't good enough (which was untrue). After I left, the customer came back and drug the party out another six months then decided to outsource its mail services to IBM Global Services.

Here's the lesson: If it doesn't work out of the box, you won't win the business. The product has to meet the corporate need. That also means, know the customer need when you come out with a product.

Linux Corporate Desktop Needs

The Linux desktop firms have done hundreds of corporate pilots for big organizations. At the same time organizations have done hundreds of pilots about which the Linux desktop firms know nothing. Here's one example of a failed pilot. A Fortune 100 company decided they wanted to go with Linux. They assembled a project team and went out and did everything they could to analyze the market and see if the desktop would work.

Ultimately, they found the product they wanted. But the company offering the product could not pass vendor requirements. Now, the desktop effort continues but they want the product to work out of the box and they want a significant supplier.

I won't divulge the company, but I will combine their needs with those of similar organizations and provide a list of functionality. Here's a short list:

That's the short list, here's the strategic list:

Small-Office / Home-Office Networking

One difficulty with Linux out of the box is setting up a small network. Many Linux users want to set up small networks that allow them to share printers, scanners, files, Internet connections and bring up a remote desktop without having to move to a different machine. A current trend includes having the ability to use a laptop or portable computer and move comfortably around an office or home with wireless connectivity. This is a plug and play small network requirement.

When small business workgroups buy PC's they tend to purchase a few name brand PCs and wire them together perhaps with a number of existing Windows machines; they then plug in a networked printer, possibly a scanner too, and they're done. Absent such simple peer-to-peer facilities, Linux can hope for practically no capture of this important market segment.

A small-business networking solution would permit Linux boxes to be networked together without a server and in short order make the system relevant to a large number of users. The SO/HO segment does not appear to have been originally specified as a Linux target; however, it's a large, fast-growing and important segment in itself and for its indirect upstream influences on the main target enterprise segments.

This is a market segment that Microsoft serves effectively and which helped that company as it moved into prominence in the late 1980's and early 1990's. Any system seeking to gain market share needs to address the basic requirements of the Small Office / Home Office demographic.

Printing

The Linux printing system works poorly and requires seasoned system administrators to set it up and maintain it. It does not work with all applications. Problems exist when printing from a browser, a wine application or when trying to share a printer.

Laptop Support

Broader WiFi card support needs to be introduced to Linux. WiFi card support for the large and important group of laptop users hardly exists. The expedient solution here would to use something like Linuxant's DriverLoader which has the elegance of being a single point solution that's applicable to the great majority of user/device scenarios.

Device Compatibility

Linux would benefit from the equivalent of Microsoft Windows Device Manager. This will provide a visual display of all hardware devices in the Linux PC and show the user which devices work and which need a module to be configured.

We suggest for unsupported devices in device manager that the user be enabled to download a driver by simply pointing and clicking on the respective device as it is listed in the device manager. Alternatively the system could provide a message that a request has been made for the device driver, with instruction on how to follow up that request in the future.

Uniformity of Different Distributions

A place exists for distributions that have no interest in the commercial market. I use one of those distributions, Debian to test, learn and enjoy pure Linux. I use commercial Linux distributions for production purposes.

If the major Linux desktop companies plan to succeed in their endeavors, they will have to learn to work together or die. Sun Microsystems just learned that they will have to delay project Janus.

Project Janus is Sun's initiative which allows Linux applications to run on Solaris unmodified. The project team discovered that Linux software has differing dependencies built-in depending on which distribution the software is built. Each software program requires separate development work for each version.

Corporate desktop buyers won't put up with this craziness. They don't want to maintain full time staffs of developers to modify software. If they did, they might as well go back to using IBM Mainframes. Except for the hassle of programming staffs, they're less expensive today that hordes of PCs.

Hurry Up

The Window will close in the not too distant future on the Linux desktop – no pun intended. The Linux desktop could fail if companies continue to pilot programs and conclude that it's less trouble to buy Microsoft. Everyone loses in that scenario.

The Linux Desktop Future

I won't use Microsoft. But, then people will soon say that I wrote the book on Essential Linux System Administration. For me and lots of others, Linux will continue as our desktop of choice.

The Linux desktop could grab major market share if the desktop companies do what the server companies did after the D.H. Brown report in 1999. If they don't make Linux a commercial competitor, then we will continue to maintain about six percent market share and a great opportunity will go by.

We will know, because the next major releases of the Linux desktop should hit at the end of the first quarter of 2005. In the mean time, some Linux desktops should sell between now and then. But who will buy them?