The Trials of a Windows 7 Upgrade

Wayne Rash
My company is like a lot of small businesses when it comes to moving my enterprise to something new, like Windows 7. There are a lot of reasons why I should, like less maintenance, better security and better reliability. But there are also plenty of excuses to delay the change as long as possible. For one thing, it costs money. For another, it means doing something about my hardware environment.

The hardware environment here is also like what you find in a lot of small businesses. While there are a couple of fairly new (meaning less than 5 years old) computers around, most of the stuff here is still running Windows XP. Worse, these machines are running XP slowly, they're not as reliable as they should be, and I spend an insane amount of time doing tech support for my own company.

And, of course, I have the problem of inertia. Doing the upgrade to Windows 7 would be time consuming. I could plan on spending a weekend getting just the computers that are used on a daily basis running the new OS. But because of an unfortunate incident, I finally decided to make the investment in time and money. It was time to take my own advice and do something about my Windows 7 problem.

What happened is that a machine performing important work suffered a drive failure. The problem was something in the storage subsystem, but I couldn't tell if it was the RAID controller or one of the drives. All I could tell is that it didn't boot. Time to make the change.

I had a choice of whether I wanted to spend an entire weekend solving the problem, plus potentially several hundred dollars in new parts for a machine that's 7 years old, or follow my own advice and get a replacement with Windows 7. I chose to follow my own advice. I gave the computer away, and went to my favorite supplier of high-end workstations, which is the Hewlett-Packard refurbished equipment list where I could pick up a machine with Vista and a Windows 7 license for less than I could buy new disks and a new controller for the old machine.

Meanwhile, I bought enough Windows 7 licenses for the machines in the company that could be upgraded (this is a very small number) and set up a replacement plan to recycle the old stuff. I already back up everything in the company that contains actual data, so this is really just a matter of swapping out machines.

The process is already under way. I set up a plan to run Microsoft's Windows 7 Upgrade Advisor on each Windows machine in the office and machines that either passed, or would pass with only minor effort and expense (like adding more memory), were flagged for the upgrade. The rest were put on a list for replacement. Of course, the concept of 'minor' depends on the machine involved. A primary dual-Xeon workstation that's already running Windows Vista needed a new SATA RAID controller. It might cost $350 to get the new Adaptec controller that would work, but it was a lot less expensive than buying a new workstation that might cost five to 10 times as much. An IBM ThinkPad that's 5 years old and needs drivers that aren't available thanks to Intel will be replaced.

The entire process will probably take a month, and I expect to make multiple visits to HP's refurb list as well as other similar lists (some of these lists are better than others). And I'll spend a lot of weekend time testing, upgrading or replacing computers. But in the end, I'll have far fewer problems, far less tech support and a lot better security once it's done. And then it'll be worth it.

Add Comment      Leave a comment on this blog post
Jan 29, 2010 12:01 AM Anonymous Anonymous  says:
Rash was rash indeed. He is presented with a relatively rare opportunity to entirely lose Microsoft, and instead waltzed out onto deeper quick-sand. What's unclear is what was being hosted on the failing disk subsystem. Ordinary NTFS files can be backed up using tar and then restored to a new system running a Linux distro (Novell SLES v10 is good and cheap) configured for logical volumes and a RAID5 disk array. If what was hosted was M$-SQL stuff, the problem is greater but not insuperable. There are multiple ways to go: DB2 on AS/400 (best), DB2 on Linux, and MySQL on Linux. Or even PostgreSQL on Linux. SQL DUMPs can create command and data streams that can then be used to totally rebuild databases and tables on a more-reliable platform. "Worse, these machines are running XP slowly, they�re not as reliable as they should be, and I spend an insane amount of time doing tech support for my own company." This sounds like the infamous M$ registry-rot problem, which may be handled by wiping C: and reinstalling Win/XP. Easy enough, granted that the sysAdmin was smart enough to initially sequester all company data and functions on D: drives or partitions, or, even better, on servers. Reply

Post a comment





(Maximum characters: 1200). You have 1200 characters left.




Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.