April 21, 2005

LWCE Toronto: Day 3

Author: David 'cdlu' Graham

The third and final day of Toronto's LinuxWorld 2005 had the meat I was looking for. First, I attended Mark S. A. Smith's presentation entitled "Linux in the Boardroom: An executive briefing". Next, I listened to David Senf of IDC discuss the top 10 CIO concerns with open source. And finally, I wrapped my attendance to this year's LinuxWorld Toronto with another session by the energetic Marcel Gagné in a presentation entitled "Linux Culture Shock"

How do we combat executive blinkers?

Smith started his discussion by making the point that he had never crashed his kernel in Linux, and while some applications had crashed, the system did not go down with them. There are fewer stability issues with Linux than there are with Windows, he said.

He described Linux as a disruptive technology; it changes everything. In two years, he said, Linux has gone from a "pretty neat" operating system to becoming a very useful and important player.

Smith cautioned about some of the risks of using proprietary software, citing an example of a city's municipal by-laws. If the city stores those by-laws in a proprietary format and, years later, it wants to open them again, the reader for that file type may no longer be licensed at the city and the city may have no legal way of opening their own by-laws.

He noted that Linux is not a great option for experienced Windows power users, while it is an increasingly good option for the casual user with some Windows experience. This due to the advent of Linux systems with a lot of Windows behaviour included.

Where Linux is most useful, he said, is on the updating and upgrading of aging systems like NT5, NetWare, MS Exchange, and so forth. From that perspective, Novell was wise to buy SuSE, as its NetWare products were getting old and a Linux upgrade made sense for the entire company. Linux also makes sense for new application installations, such as a new database deployment, and where security is a concern.

On the topic of transition costs, he noted that Oracle, among other large software vendors, licenses their software by the number of processors it is running on, not the operating system. Thus a license for 4 processors in a traditional Unix being transferred to Linux would not cost any additional licensing fees, just a trade of software files.

He commented that a wholesale replacement of Windows systems not otherwise ready for upgrade to Linux was comparable to trading an entire fleet of trucks in for a fresh fleet of trucks -- the economic sense is not there.

Of all the major industries that use computers today, he said, there are none that have not already started using Linux.

What executives are looking for in their systems is industrial strength systems with modest customisation, a longer life cycle, and stability. They want to avoid, argued Smith, non-delivery of software or specifications, systems that are difficult to support, high training requirements, paying for incremental increases in users, and any kind of negative impact on customers.

Linux upgrades make sense, he said, for companies that are in the up-swing of their business cycles. A company in a downturn would do well to stay with what they are using until they're on more stable financial footing.

Linux delivers low cost stability, security, improved performance on the same hardware, flexibility, business agility, a natural upgrade path, easy of use, and consistency. It reduces costs, he said, but that is more a bonus than a rationale for switching to Linux. He warned that focusing on the cost savings of switching to Linux could negate them as people try to improve them artificially.

Linux' strengths include rebootless upgrades and high uptimes, and very little need to reboot. He noted that there have been many a Linux server that have simply not been rebooted in the space of more than a year.

Older systems support Linux with little trouble where Windows would have trouble running, reducing hardware costs. It requires less RAM than its proprietary counterparts to accomplish the same tasks, and requires fewer administrators to administer large numbers of Linux systems.

In order for executives to be sold on a switch to Linux, they have to accept the factors he discussed and see the return on investment possibilities, largely from savings from more efficient hardware, software, and use of employee time.

The Gravity of Open Source

David Senf of IDC led a session entitled "Gravity of Open Source: Top 10 CIO concerns" in which he listed, in reverse order, the top 10 concerns of chief information officers related to the adoption of Linux and open source, based on studies conducted by IDC. The studies seek to find the questions CIOs have, but it is left as an exercise to vendors and others with an interest in deploying open source and Linux to actually provide the answers.

10. Intellectual property concerns extending out of the SCO vs The World lawsuits came in at number 10.

According to their survey, only 3% of survey respondents indicated IP issues had stopped them from implementing open source solutions, though 54% said it wasn't applicable because they weren't looking at converting to Linux or Open Source anyway.

9. What open source business models will succeed?

CIOs have expressed concern about choosing software packages or applications that will no longer be developed or supported at some point in the future. Which software will last? Sticking to larger, better known companies such as IBM, Novell, Sun, HP, and Red Hat, is, for many, the answer. These companies are driving the agenda for open source development and are useful to follow.

Senf discussed SuSE and Red Hat Linux distributions and the effect they are having on each other. He predicts that over time, the two will balance each other out and have a roughly even market share.

8. Where is open source going?

Is Linux the end point, or the means to something further? Organisations are more likely to deploy open source given Linux' success.

7. What workloads is Linux good for?

Over the last year, Senf said, Linux has grown by 35% on servers. Its main use is in information technology departments and functions, and as a basis for web infrastructure. Linux is becoming a powerhouse in high performance computing.

6. Can we have too much open source software at once?

How much open source software should we have?

There's an increase in open source use of variants of SQL database systems, and Apache is the dominant web server, among other examples. What software and how much software is needed, however, is a matter of discretion and varies wildly from one company to the next.

Who is using the software is also an important factor. Senf noted that knowledge workers tend to need open source software less because of the specific tasks they are doing, while transactional workers are more flexible and have more use for the open source systems.

Open source software can also be implemented in stages. For an example, he pointed to OpenOffice.org and FireFox which are both run primarily on Windows systems for the time being.

5. What channels are being used to acquire Linux and open source software?

Open source software other than the Linux operating system itself is generally purchased through solutions vendors, while Linux is generally acquired through normal public distribution channels such as downloads or inexpensive CDs.

4. Why are organisations and companies adopting Linux and open source software?

For Linux in the business environment:

The number one reason given is for the cost savings. Second is that it gives companies an alternative and leverage against software providers. If they are under-performing, they have an option for switching. Third is improved return on investment.

For open source software, from a business perspective, the first two reasons are the same, but the third is protection from vendor lock-in. Senf indicated that he disputes this, noting that it is no less complicated or expensive to convert from open source packages than it is from closed to open source packages.

His own rationale is that Linux deployment offers improvements in speed, performance, and security. Open source software offers improvements by offering functionality not available in closed source software. Secondly, it is user-modifiable. Senf noted that while many companies offer this as a reason for switching to open source software, he doubts many companies actually do the modifications they are provided the right to do. Finally, security is offered as the third reason for deploying open source software in businesses.

3. What is up with the total cost of ownership debate?

CIOs, on average, estimate, according to a chart Senf provided in his presentation, that training, integration, internal and external support, and installation and deployment all cost more under Linux, but that administration, downtime, and acquisition costs are comparable or cheaper with Linux than under proprietary operating systems. How this comes out at the end will vary for each company and implementation.

2. Why are organisations and companies not adopting Linux and open source software?

1. How well does it fit in with our business requirements?

CIO concerns in this area include productivity questions, existing markets, and the maintenance of customer service, as areas that cannot be compromised for the adoption of Linux and open source.

His list finished, Senf noted that with the advent of big-vendor support for Linux, the risks of open source software are, in his words, taken out of the game.

Linux Culture Shock

I wrapped my attendance to this year's LinuxWorld Toronto with another session by the energetic Marcel Gagné in a presentation entitled "Linux Culture Shock". Gagné started his discussion with a comment about avoiding playing with the Linux kernel. He said it is no longer necessary to tweak or compile a Linux kernel. The stock kernels are more than adequate.

He touched on the fear of spending money for retraining and the myth of Linux desktops not being useful. In an effort to show how this is indeed a myth, he began a quick tour of the Linux desktop under KDE, right clicking on his desktop to show the configuration menu and other features familiar to users of Windows.

Mozilla FireFox, he said, is not perfect software, but it is more perfect than the proprietary Internet Explorer. He reminded the audience that the Department of Homeland Security recommended against running IE, because it is so insecure. OpenOffice.org 2.0 (still in beta), more than its predecessor versions, is very close to Microsoft Office. And Gaim, he said, is a good option for instant messaging solutions for users transitioning from Windows.

Regarding Gaim, an audience member asked about deliberate protocol breakage on the part of instant messaging service providers. "The open source community is nothing if not resourceful" was his answer. Some time ago, he related, Microsoft broke the MSN protocol to break non-authorised clients. It took a couple of days, but the open source developers were able to reverse engineer the protocol and restore the open source messaging client's ability to communicate via the service.

Unlike ICQ, AIM, MSN, and Yahoo instant messenger, Gagné noted, the Jabber protocol is an IETF (Internet Engineering Task Force) standardised protocol. Both client and server software is available for the Jabber protocol from the open source community, from jabber.org.

Gagné opened the floor to the audience to tell of things they are hearing in the way of resistance to switching to Linux, and offered responses to the various concerns. The first question was on the easy -- or not so easy -- installation of video card drivers for Linux.

He answered that this problem is no longer a serious issue and hasn't been for several years. With the exception of some of the latest bleeding edge video cards, while they are still bleeding edge, most video cards are supported properly in Linux. All video cards that he is aware of in the last several years support the VESA standard and should at minimum work under Linux.

Question number two was how to address the issues of total cost of ownership and support. Support is the question of "If something goes wrong, who do I call?" he said. With large companies supporting Linux, they will be willing to sign support service contracts with companies that want them. IRC channels and mailing lists are the perennial free support option, and are often faster and certainly cheaper than any Microsoft support requests.

From a total cost of ownership perspective, he said, Linux requires fewer people to manage it and has fewer problems in operation. Because of Linux' networking roots, it is possible to fix a lot of problems remotely and the cost of travel to fix broken systems is dramatically reduced with the use of Linux.

Problems like email viruses cannot be discounted from total cost of ownership calculations, he pointed out, noting the ILOVEYOU mail virus some time ago which brought hospital computer systems, among others, to their knees for days as they tried to recover.

Why, an attendee asked, are there no viruses for Linux? While noting that some viruses had been written in perfect circumstances as proof of concept cases, Gagné discussed the history of Windows versus the history of Linux, with the latter having been developed on the Internet in an environment where security is a concern. Windows, on the other hand, was never initially developed with security in mind.

Windows' tendency to overuse the administrative accounts and its memory sharing policies increase its vulnerability to security issues like viruses. He cited a honeypot project report which suggested Windows systems, out of the box, put on the Internet had a life expectancy of about 4 minutes, while a Linux box put onto the Internet could be expected to last about 3 months before being compromised.

A discussion ensued in the room about the possibility of viruses becoming a problem on Linux should it become the dominant operating system. While Gagné acknowledged the possibility, he described the logic of not upgrading to Linux because of that possibility as strange. Why, he asked, would anyone refuse to upgrade from an obviously insecure system to one that might potentially become so later because of that potential?

The next question was about training. His answer to the problem of training users to use Linux desktops is simple. Sit down with it and use it. He described the transition from Windows to Linux as no more of a culture shock than the transition from Windows 98 to Windows XP. Both require some new learning and getting used to.

And that's a wrap

LinuxWorld Toronto is -- as of this year -- a member of the LinuxWorld show family, rather than an independent operation as it had been in previous years when it was known as RealWorld Linux. It was a fairly good show. The second day felt a little marketing droidy, but it is difficult for show managers to know in advance what the quality and nature of presentations will be. The first and third day, and some of the second day sessions, showed that a good focus could be kept, and that the show can be useful for Linux' continued growth.

Click Here!