Taking a Bott beating

My colleague Ed Bott, who writes the Windows column for ZDNet, takes issue with the problem solving approach that I used when I went about migrating my friend Christine's old PC over to Vista. From the nature of the talkbacks to that original post, I would say a number of you do as well.

Click on the "Read the rest of this entry" link below for more.

With all due respect to my colleague Mr Bott -- who is someone for which I have immense professional regard in what he has achieved in his career as a technology writer and who's hands-on Windows Kung Fu is currently superior to my own, I would like to state that the solution that he proposes is "Old School" and what I attempted and successfully executed with my friend is "New School" and is a technique that will become increasingly common, particularly in large enterprises which template XP installations from master install images.

I maintain that the approach I used is less invasive, is inherently safer when the correct caution is undertaken, and has more potential for end-user satisfaction than the traditional PC migration procedure that he proposes. I will also note that virtualization of the existing legacy environment is the only guaranteed and sure fire way to ensure complete end-user business continuity from one system to the next. I also believe that this  should be a major design element for the next version of Windows -- and Bott has also disagreed with me as well on this issue so naturally his response can be attributed to his belief that end-user virtualization for PC migrations is overkill.

As a Systems Architect, integrator and a virtualization specialist, my world view and use of computing tools and techniques are different than Bott's, who is more end-user focused. The methodology and solution approach that Bott proposes is valid, and it will work, but it's not without its own pitfalls.

First is the issue of data transfers. Bott as well as some of you propose that attaching a USB hard disk on the old PC and using a system recovery boot CD to copy the data "out of band" is slower and more painful than simply yanking the original disk out and attaching it to the newer, faster PC by using a SATA to IDE interface dongle or some sort of external housing. Yes, my method is certainly slower, and that's a limitation of USB 1.1's vastly more reduced transfer rate compared to USB 2.0 or a direct SATA connection.

However I will note that I had a very good reason for doing what I was doing, and that was to be as LEAST INVASIVE AS POSSIBLE. As someone who has spent 20 years in this industry, and a significant amount of my early career cracking open tons of PCs and servers and minicomputers as a field service tech in Fortune 500 companies in every work environment possible, from the biggest banks on Wall Street to industrial manufacturing clients, I can tell you from experience that there is always the chance that with an older system, the inside of the chassis and the mainboard and other critical components will be caked on with dust and other particulates, especially if it has been shoved under someone's desk for several years and has never been vacuumed out or cleaned in any way. If you crack open the case and yank that hard disk, you will run the chance of electrostatic discharge and damaging the hard drive, even if you use a wrist collar.

In the case of this particular PC -- of which you can now look inside -- it was in the house of someone who was a smoker, that smoked near their PC when they used it, and the inside of the chassis and the hard disk was caked on with dust bunnies and ash. While you could blow the dust bunnies off with an air blower, you again risk a discharge and damaging system components that you may need access to later. On newer systems that have been adequately ventilated and are used in relatively dust-free environments Bott's technique is valid. That's fine for an ideal world. It is much less valid when the system itself is at risk, and I knew just how risky it was the second I eyeballed the machine for the first time.

Bott also brings up the subject of interviews and finding out where all the data is stored and simply copying that data off.  This I did do, and Christine now has all her Office documents, digital photos and her Outlook data imported. However, Bott's solution does not address applications for which the installation media does not exist or are simply incompatible with Vista. He is also completely ignoring scenarios where certain critical applications are "abandonware" and/or are not certified to run on the new OS by the originating vendor. In Christine's case, she actually had a valid business need to continue to use Office 2003 and a number of other applications which would make no sense to install in Vista.

Christine herself purchased an Office 2007 license, but in some cases, a user might not want to go out and spend several hundred dollars on a new office suite immediately, and despite the existence of Free and Open Source productivity suites such as OpenOffice and Lotus Symphony, compatibility is not yet completely baked. In this case, a P2V virtualization conversion of their old PC with all their applications and data and settings in place is not only prudent, but the only sure-fire away to know ALL of the data that the end-user needs is copied over. Christine told me about her major directories, but it's hard to tell sometimes if some obscure data file was dropped six directory levels deep under My Documents or in some unconventional location. The user may need to locate it later and move it to the new system. Giving them a USB dongle to their old hard drive is one way to go about it, but if the data is unusable without the original application, its not going to help much.

Virtualization to solve end-user migration problems may sound like overkill for anything but the most demanding enterprises today, but it will eventually become the de-facto way of addressing these problems in the near future even for small businesses and home users. This you can count on, as well as home and end-user virtualization/migration solutions from the usual suspects that will make my labor intensive cleansing and P2V process eventually become as easy as booting a CD and attaching a USB storage device.

So to recap, I believe both Bott is correct and I am correct, but the importance in fully understanding and addressing an end-user's needs is paramount to the solution approach used. But ultimately, I believe virtualization is the only way to ensure a complete and painless migration for the end-user. It may currently be "a headache" or "overkill" for the practitioner but anything worth doing requires effort.

Do you agree or disagree with my approach? Talk back and let me know.