Migration

Digg This!

In my editorial in last month's LinuxWorld Magazine, I defined inflection points (with the help of Intel's Andy Grove) as those things that change our behavior with regards to our IT strategy. I was speaking of software and hardware upgrades and other realities of running a business that make us consider other options before investing in our IT infrastructure. Typically, this means purchasing newer versions of the same solutions we've been using.

I believe many of these proceedings are on the horizon, but the question is, how deep is that horizon? For example, every year since 2000, I've heard some industry pundit, analyst, or Linux evangelist talk about it being "The Year of the Linux Desktop," and to the disappointment of many, that has not yet happened. While the adoption of Linux as a desktop platform is growing, there is no mass migration to Linux for the productivity desktop. That's not to say that there aren't a substantial number of Linux desktop users. To cite Google Zeitgeist, May 2004, (www.google.com/press/zeitgeist.html) only 1% of the operating systems visiting Google were running Linux. In contrast, 91% of the operating systems accessing Google during the month of May were running a version of Microsoft Windows. That's quite a gap, but when you consider the speed at which Linux has garnered that 1%, you have to wonder how long it will be before it overtakes Mac at 3%. After all, the Mac has been around since 1984, and according to this Google reporting, it still only controls a small part of the market (see Figure 1).

For Linux to acquire greater market share, something has to make Linux more compelling as a desktop OS. That's why I have chosen to predict, or at least identify, trends that may move us closer to widespread Linux desktop use, or at least signal Linux as an even better opportunity than it is today. Early Linux desktop adopters may find that adoption in a configuration that's slightly different from the way they're using PCs today would serve them well in either a Linux or a Windows world.

Personal Data Storage in the Network

Because the migration of operating systems often includes the migration of data, it may be time to rethink where we store the data. If you are going to migrate to Linux and have to move your data anyway, maybe it's time to consider storing that data in the network. That way, whenever you change PCs or need to be mobile, your data is accessible.

Today, we mostly store our personal data on PCs, which may or may not be backed up to the network. Over the long term, you may not store data locally, which is a tough pill for many people to swallow –the idea of your data being somewhere at the end of a wire that you have no control of is pretty daunting. After all, your data may contain not only financial records but also personal correspondence, pictures from a family vacation, and other precious items that, if lost, would not only cause you great inconvenience but could expose you to other dangers, such as identity theft. However, there are services out there that I would never have considered 10 years ago that today make more sense.

One of the more sensational services is Google's Gmail (http://gmail.google.com) e-mail service, which allows you to store up to 1GB of e-mail on Google's servers in exchange for your allowing them to advertise to you based on the content of your e-mail. This raises many privacy concerns. I don't believe it is Google's intent to look into your private life and supply that information to the world, but whenever a mechanism like that exists, so does the opportunity for that system to be compromised or abused. I have been playing with my Gmail account for about a week now and find it to be very useful; I would also consider paying for such a service if it included a service-level guarantee and amenities like SPAM filtering. Interestingly, Google seems to be banking on the advertising revenues generated by this service to subsidize the costs if not make it a profit center. More to the point, leaving your e-mail and other data in someone else's hands is analogous to leaving your money in the bank; you need to have some level of trust in the validity and security of the system for you to leave your money in the hands of a stranger, and the same trust must exist for you to do the same with your data. Once your data is in a position to be accessed over a network, that sets the stage for it to be accessed by application service providers (ASPs), which is when your daily computing environment could really change, whether it's on Windows or Linux.

Greater Availability of Robust Applications

There are a great number of applications available for Linux, and there's an adequate software package for about 80% of all productivity computing tasks. For widespread Linux adoption, the remaining 20% of the functionality needs to be addressed. In the long term, that need will be addressed by one or both of the following methods; in the short term, bridging strategies will be necessary, as discussed later in this article.

Applications Become Web Services

Far-reaching application availability may not come in the same package we are familiar with today. In a recent article alluded to in Red Hat's Under the Brim newsletter, Red Hat chief Matthew Szulik talked about how his vision was that, in five years, your calendar, e-mail, and word processor will become Web services and the infrastructure for those Web services will be provided by open source technologies. While I whole-heartedly agree that his vision is correct in the very long term, I suspect that day is farther out than five years. Perhaps one of the early indicators that Web services have a future is our growing dependency on the Internet for content like news, driving directions, and banking. The evolution of content provided through this medium could be productivity applications, but for that to happen, we need to have pervasive Internet access. Inklings of this are starting to emerge with the availability of data services on our mobile phones and wireless hot spots at coffee houses around the globe, but omnipresent access to rich content through wireless networks is still not a reality in much of the world. My point is this: some of your critical applications for Linux desktop adoption may exist today but they exist in the form of Web services. For example, TurboTax (www.turbotax.com) offers a Web version of their software so that you can be platform independent in using their services. Perhaps some of your other key applications will make their way to the Web as well, making your productivity desktop less platform dependent. With that dependency solved, Linux becomes a more attractive choice, especially when you look at the administration and software licensing advantages.

Major ISVs Developing Native Linux Applications

The Web services vision for Linux applications is fairly futuristic and we've spent the past 20 years becoming endeared to our PCs, so I do believe that there will always be a need for applications running locally on your PC. Trusting applications and data to someone else requires a huge leap of faith. That's why people are (understandably)hesitant when adopting Linux for use in personal finance and accounting. As one accounting software manufacturer recently told me, once someone adopts an accounting package, they rarely ever change it just because they are putting a lot at risk by moving, as well as being inconvenienced by downtime and data migration. However, these same "bean counters" are also well aware of the costs of their IT infrastructure, and given an opportunity for considerably better Total Cost of Ownership (TCO) of IT infrastructure, they are apt to evaluate applications that run natively on Linux.

Widespread enterprise implementations of desktop Linux will most likely have to pass through the CFO's office, and many CFOs will want to know if their applications run on the same platform as everyone else's. That's why financial applications for Linux are going to be a consideration for desktop adoption even if for more political than practical reasons. This may be a seemingly insignificant part of the Linux desktop equation, but sometimes Linux adoption will be driven not only by facts but also by perception. It's not uncommon for me to hear Linux diehards comment that they normally use Linux but they still have one Windows PC with Quicken and/or TurboTax to supplement their Linux desktop. That's because there is no Linux application that instills the widespread confidence to handle their fiscal information. But financial packages are just the tip of the iceberg; when a wide variety of applications that are easily installed and supported become available to solve that last 20% of user needs, then we will really begin to see increased desktop adoption.

Expertise and Vendor-Neutral Management Tools

Once Linux is installed on a PC, performance is usually excellent and in my experience downtime is rare. Very rare, that is, until a change occurs; that change could be an update to the OS kernel or the addition of new hardware. Increases in productivity due to less downtime can be quickly eaten up by administration overhead.

It's not adequate to just have tools and solutions that can get the get the job done; it's more important for those tools to be leveraged by existing personnel who can quickly gain the expertise to administer Linux systems. The learning curve needs to be minimized wherever possible. Intuitive, easy-to-learn tools are an essential part of successful migration strategies.

An emerging Linux software company, Open Country (www.opencountry.com), spoke to me about their vision for managing Linux deployments. Since Linux is an open system they want to provide the tools to manage the network. Open Country's COO, Laurent Gharda, says, "Providing an easy-to-use system management solution is of great importance to deploy and manage Linux for those companies that have little Linux expertise or expertise in another operating system." Not only is it important for that tool to exist, but for it to be vendor neutral. Mike Grove, Open Country's CEO, adds "We believe being vendor neutral is important so that customers aren't locked into one distribution over another." They believe that their OC-Manager management console will allow administrators to overcome the two underserved items when it comes to Linux system administration: management complexity and software unpredictability. Their success remains to be seen, but as more vendors concentrate on the usability aspects of the use and administration of the Linux desktop, uptake in Linux adoption will increase.

A company thinking along the same lines (though not in terms of vendor neutrality) is Xandros (www.xandros.com), makers of the popular Xandros Business Desktop. They have announced the development of the Xandros Desktop Management Server, which is intended to allow for deployment and management of Xandros desktops. While their solution is aimed at Xandros desktops, it's another example of mass deployment tools that will be developed to try to aid adoption in the enterprise. All of these early efforts are worth watching, as they could be potential aids to your successful migration to a Linux desktop platform.

Breaking Vendor Dependencies

It's time for you to not only re-evaluate operating systems, but also to evaluate new technologies from a platform-neutral perspective. I don't expect anyone to toss their existing infrastructure in favor of a brand new OS; costs of migration right off the bat could be more substantial than using an incumbent solution. However, if you are evaluating a new technology, it's wise to consider whether it will lock you into a dependency on the operating system or browser for Web applications. Even if you don't adopt Linux today, you should consider how purchasing something today will affect your choices tomorrow. Any time you have options to make multiple choices on a solution, you can negotiate prices and make sure that the services and supporting features are best suited for your needs. There is nothing more frustrating than buyer's remorse; you may want to reflect on these points when you begin to evaluate a Linux migration.

In the Meantime: Staging a Linux Migration

There are few things you can do to prepare for Linux as a desktop platform.

Build Internal Expertise

While you can't spend too much time investigating every new technology that comes on the scene, building expertise within your organization in preparation for serious Linux evaluations is a good investment. It's a good investment for two reasons: you may realize the cost savings Linux can provide you, or it may save you from migrating to Linux in a way that costs you more than it saves. Additionally, this will help you understand what would make good Linux pilots and will help you make informed decisions when it's time to come to a verdict on Linux as a solution. Plus, Linux server expertise, which is a very good investment today, should translate to the desktop when you make that commitment tomorrow.

Develop a Bridging Strategy

Identify solutions that help bridge the gap between ideal Linux solutions and what's workable today. These bridges would include any of the following:

• Cross-platform applications: Areas for evaluation would include office suites that can be implemented on existing systems before a large scale OS migration, e.g., Mozilla Web browsers (www.mozilla.org) and Sun's Star Office (www.sun.com/staroffice).
• Emulators and virtual machines: Recognize those absolutely essential applications you use today in your Windows environment and look at the potential of migrating them intact to Linux. There are a variety of ways to do this via a virtual machine or emulated environment in which the Windows application is executed as intended, but in a virtual machine running on Linux or through an emulated Windows API. The solutions to consider for this include VMWare (www.vmware.com), Win4Lin (www.win4lin.com), and Wine (www.winehq.com).
• Terminal servers: Another approach to filling the gap between widespread application availability and the current situation is to host and redisplay critical applications via a terminal server and redisplay them to Linux desktops. Applications can even be running on Microsoft Windows servers via Windows Terminal Server and redisplayed to Linux desktops via solutions like Citrix (www.citrix.com), Tarantella (www.tarantella.com), and GraphOn (www.graphon.com). However, this is probably the least attractive way to migrate because it requires a mixed operating system footprint.

Evaluate Your Existing Infrastructure

Take inventory of what is going on in your enterprise today. Develop metrics that give you a snapshot of how events like PC downtime affect your bottom line.

• Look at actual costs of IT systems: Cost involves a lot more than software licensing and hardware. There are costs associated with viruses, spam, and the resulting downtime and administrative overhead to deal with them. Particularly, evaluate the potential for open source solutions to reduce these costs. In this scenario, it would be wise to look at any solutions that are subject to viruses or require costly closed source software to maintain.
• Determine usable hardware life: The city of Munich, Germany, has received a lot of media attention since announcing their intention to move to Linux as a desktop platform. One of the deciding factors for this was the costs associated with upgrading their hardware to support newer versions of the Windows operating system. Their problems aren't unique, and you may find that users' computing needs aren't changing as fast as hardware is being improved or software versions are coming out. Decide if upgrading to the latest and greatest is going to offer more value than older hardware and a Linux operating system.
• Determine user needs: I have this dilemma every year when new cars come out. I often think how nice it would be to have the latest model equipped with the latest and greatest gadgets. However, when I evaluate my needs versus my wants, I find that very seldom can I find justification for such a purchase. IT equipment is very much the same; even though faster processors, newer operating systems, and and newer applications have been released, the needs of the end user are probably very similar to what they where when older equipment was first purchased. If end of support life is a factor in your decision, make sure that you don't overbuy during your technology refresh.

Conclusion

I would like to say with 100% confidence that anyone reading this article could make the move to Linux on the desktop today. However, that wouldn't be true –there are many factors limiting the adoption of Linux, but those factors are becoming fewer as the days go by. My recommendation is that, if you are interested in a Linux enterprise and you have objections that can't be overcome today, that you watch as more milestones are reached. Also, talk to vendors you may already trust who have a Linux strategy today (Novell, IBM, and Sun come to mind). Then choose the one whose products are most complementary to your objectives and start to formulate a bridging strategy to successfully migrate from one platform to another.

I advise this much time in evaluating one technology largely because open source initiatives are probably the most "disruptive technologies" that have the grandest potential to be successful at all levels of computing use. In the upcoming years, Linux advancement in both server and desktop applications could have the potential to help control one of your costly expenditures as well as enable you to have more freedom in your technology environments. By increasing productivity and providing competition in the desktop computing arena, there can be nothing but good things to come.

Tomorrow's Linux desktop may focus less on the client and more on the network. The ecosystem shown in Figure 1 would be conducive to Linux as an end-user platform.

-SIDEBAR-

Linux Migration Success Stories

Both LilleCorp and Addison UK are having tremendous success promoting Linux on the desktop, they are doing so by following an easier-to-manage thin-client strategy.

LilleCorp
Jordan Rosen, CEO of LilleCorp (www.lillecorp.com), a systems integrator in Albany, New York, has identified that the healthcare industry is a prime candidate for Linux solutions. They have proven that thin-client Linux solutions are helping Capital Cardiology Associates, a medical provider utilizing Linux IT infrastructure with many benefits. Their solution is to provide thin-client computing architecture powered by Linux that minimizes cost through reducing the need for IT services and, more important, lost productivity due to computer downtime. Their network-based solution allows computing environments to be portable while still having a single tightly controlled point of deployment. This high level of control allows them to maintain integrity of the operating system, minimizing downtime caused by user errors. Additionally, they profit from the fact that the Linux OS is somewhat less affected by viruses due to the disparity in the number of viruses that attack Windows systems versus Linux.

SchoolLinux
SchoolLINUX.com (www.schoollinux.com) is a division of Addison UK, a developer of Linux solutions for the education and small business. They focus on creating easy-to-use servers to provide back-office infrastructure and desktops that stand up to heavy multi-user traffic, which is typical in a school with a shared workstation environment. Their approach is to provide easy-to-use ICT technologies to schools in the UK that are both cost effective and useful. Their Linux-powered solutions are all ideal solutions for schools who before were dependent on Windows solutions only.