When Edward Haletky's friend asked him for help setting up a Linux desktop in the year 2000, they found only half of the Web applications needed. Since then, while researching his new book, Deploying Linux on the Desktop, Haletky has seen desktop Linux application availability and usability increase to the point where it's nearly ready for widespread corporate use. Yet Haletky does not think that Linux desktops will be widespread by 2007. In this interview, he explains why.
What is the current enterprise demand for desktop Linux?
Edward Haletky: The current enterprise demand for desktop Linux is growing daily and is very hard to quantify at this time. However, there are two desktop efforts going at the moment. The first is for the home user, and the second is for the enterprise. While these may seem dissimilar, they are in essence the same in most respects.
The difference boils down to either the custom enterprise applications or specialized tools to access mail and enterprise databases. But in many aspects: for information sharing and training, a good Web and connection client is all that is necessary. For information generation, a good office suite is needed. Both of these are available on Linux today. There are many things to overcome before Linux will be a primary desktop for most users.
What are some of the most important types of custom enterprise applications?
Haletky: Specialized tools could be Websites that absolutely require Internet Explorer or custom Cobol applications for accessing company databases. Or it could be as simple as a VPN tool that has no Linux equivalent yet. In essence, any application that locks you into a single operating system, tool, or computing hardware.
How does Linux stack up to Windows for handling those apps?
Haletky: Enabling tools like CrossOver Office/Wine can help here, as they will run Native windows applications on top of Linux, but unfortunately it's not 100% yet and while it is extremely useful for most applications, there are some that absolutely require the real thing.
This is changing significantly but it's not there yet. Since the list of available packages grows by tens every day, there could be one that solves your problem or accesses your data. Applications that require Internet Explorer can easily run with ASP plug-ins for Firefox or even under Wine. But those custom Cobol and other applications would take a bit more up-front testing and certification in order to switch over.
Do you think Linux will be common on desktops by 2007, or is it still going to be primarily a server distro?
Haletky: By 2007? Linux will be a server and desktop distribution. While Desktop Linux will grow to numbers higher than we can think at the moment, its strength really is server applications.
What do you see as the major barriers to using Linux on the desktop? Is application availability still an issue?
Haletky: All the Web applications exist, as does an extremely good office style suite. So the basic applications are there, and with tools like Wine (CrossOver Office) many of the windows desktop tools are also available for Linux. However, Wine does not yet support .NET and other custom applications. Support for more applications is changing daily and will be a part of the next major release of Wine.
Is it also a problem of perception?
Haletky: Perception is also a problem. With many users use to Windows and MacOS going to Linux is considered too difficult with too hard a learning curve. The perception HAS changed over the last years but there is more still to do.
Desktop-specific distributions are going a very long way toward alleviating perception problems. Lindows, RedHat Desktop, even JDS on Linux (Java Desktop System) have all been working towards this goal. Perception is a hard thing to overcome, but there is a lot of effort happening in this arena.
The other perception problem is with [Linux's] method of development. Open source is still a bit of a dark horse to many people who consider it to be without support -- which is really untrue, but it's a perception that needs to be worked on.
What are the major benefits of deploying Linux on the enterprise desktop? I've heard a lot of arguments for cost savings and security... but is there another compelling argument?
Haletky: Cost Savings may or may not be there; however, in most cases they are there! Security by and far is better with Linux than any Windows box out there. There are numerous studies to this effect. However, what I consider most beneficial are the support tools available. With Linux there is nothing hidden so it is easier to debug the problems. It is far easier to roll out changes using the distributive tools that make up Linux than anything else I have attempted. With items like dkms, kernel updates are even simplified.
The thing not available on most other desktops is the features of the kernel, which grow daily in an open source community. Tools like software RAID make a huge impact. For example, my laptop is not new, but due to software RAID, I have a mirror set in case my hard drive fails and it can use any USB drive I plug in to the box for this functionality. Software RAID can be used to cut the cost of hardware while increasing redundancy and lower the overall cost of support.
Linux also just plain feels faster than any other desktop I have used on similar hardware. With everyone doing more with less, Linux makes a very good desktop as it runs on just about everything and has features that make it more responsive.
Which distribution would you recommend for the desktop, and what makes it your choice?
Haletky: I really do not recommend one distribution over another; each has its benefits and special features. I personally use RedHat Linux but find that SuSE has a different but similar set of tools and that Mandrake ships many more desktop applications and tools than the other two. If you really want to find out how Linux works, distributions like Debian and Gentoo have their places. Yet, there are now distributions built on the main ones that contain other items like Ubuntu which is very popular these days.
What criteria would influence you to recommend one distro versus another in certain situations?
Haletky: I like a desktop that is supportable by an outside source, like RedHat or SuSE, yet I would not rule out using Debian-based distributions as they too can be supported. The Linspire distribution bundles CrossOver Office (wine) with a Linux Distribution which solves many issues for users. Yet supportability is my major factor for deciding on a distribution. Which distribution meets my requirements while maintaining supportability is the governing factor for a Corporate desktop.
With over 14000 possible packages, [a number] which grows daily, most distributions only choose around 6000 packages. Mandrake supplies more tools for multimedia than SuSE or RedHat. Yet Debian has the ability to load many more packages than any other distribution. So finding something that meets your needs could be simple. Finding something supportable is another question that Deploying Linux on the desktop tries to answer.
What are the steps an IT manager needs to take in a Linux desktop deployment?
Haletky: The most important step I often find is overlooked: to plan and get a rigorous list of requirements. Make sure you understand the costs and tools that will be implemented so you can understand the total cost for your desktop environment. Refine these requirements with interviews with real users or even watch them work. You may find that what you thought was a requirement moves into the "nice-to-have" realm. After requirements there is developing a desktop, then testing and deploying the desktop with the appropriate training.
With 16,000 possible packages and more appearing daily, one needs to be open to all the possible choices that exist with Linux.
Edward L. Haletky graduated from Purdue University in 1988 with a degree in Aeronautical and Astronautical Engineering. Since then, he has worked in programming graphics and other low-level libraries on various UNIX platforms. He currently works for Hewlett-Packard in the Linux/VMware Expert Team, dealing with Linux Clustering Technologies, as well as general Linux and VMware environments. He has also published articles on the subjects of interoperability, clustering and security for Linux.