Linux, OpenBSD, Windows Server Comparison: Reliability and Windows 2000 Blue Screen Ad

In a serious advertising blunder, at the beginning of 2001, Microsoft ran a series of ads in the trade press. Under a picture of a "Windows 95 A fatal exception . . . " or blue screen, the ad started with "GOODBYE BLUE SCREEN, HELLO RELIABLE MICROSOFT", and subsequently stated that "Windows 2000 Professional is 13 times more reliable than Windows 98." Those who work with a variety of systems learn, especially in Windows and Macintosh families of systems, that there can be an enormous amount of variability in the stability of any two systems running the same operating system. Still, the same experienced persons know that some OSs are generally more stable than others.

To put the ad in perspective, you have to ask what is the least stable OS sold in recent times. The candidates would be Windows 95, Windows 98 and Macintosh systems from the mid 90's until the recently released OS X Mach based systems. Personally, I've managed to avoid working on any of these as my own system (I went from Windows 3.1 to NT 3.51 at work and to NT 4 at home) but have talked to a variety of users and administrators. My sense is that 98 is slightly more stable than 95, unless it's upgraded over 95, in which case, 98 will be less stable. Macintosh was probably somewhat more stable than either 95 or 98, but less stable than earlier Macs. With Macs from the mid 90's and all Windows family systems, including servers, a seemingly minor change can wreck a previously decent system.

I know from where I've worked, from my brother who is a software development manager at one of the largest companies in the world, and from a friend who is a contract system administrator at the Census Bureau, that with both 95 and 98, when a system experiences problems that can't be easily fixed, that systems are provided fresh standard installs.

I believe that NT is significantly more stable than 95, 98 or Mac. Part of the reason I never used 95 was it just did not appear to be enough more stable than 3.1 to be worth the upgrade effort. In early 1996, when my 3.1 system crashed four times in one day, I ordered NT 3.51, and was generally pleased with the results. I learned that some specific applications could crash NT, just about as easily as anything crashed 3.1, but the behavior was consistent, and once I learned what to avoid, had a reasonably stable desktop system. Subsequently, I've had an NT 4 workstation up about 6 months. I've never had an NT server stay up as long.

The flip side is that I've seen infrequent blue screens on most NT systems I've used, and also had instantaneous hard lockups, where the total lack keyboard, mouse, or network response, suggested a hard crash, without even a blue screen warning. The biggest problem with all Windows systems, is the complete inability to predict when a seemingly minor system change, will cause major system problems. Ninety some percent of Windows software installs are simple and uneventful, with the desired result. Then, for no apparent reason, something will go wrong on a another install or system change.

One of my worst NT experiences was on my workstation, when I upgraded to Internet Explorer 4. The first error message I remember (it wasn't the first), was one telling me there was a problem with Internet Explorer, and to reinstall it. Since that was exactly what I was trying to do, the message was more than a little frustrating. The situation worsened and I decided to reinstall the OS and restore from backup. I hadn't yet learned to keep two working copies of NT on the boot partition for just such occasions, and did not want to install over the corrupted system, nor change the system directory name.

I deleted the C: partition, assuming the install would allow me to reassign the now empty space, and reformat. Instead, it turned the D: partition into the C:, and I could find no way to undo that. I had to repartition C: and D:, install a basic OS, and restore both C: and D: As this was my home machine and I had an automated, early A.M. backup from that morning, I lost nothing but time. The restored system had to be, for all intents and purposes, identical to the one I'd tried to upgrade IE on. I had not done any work, before I started the upgrade. The second install completed properly, without problems, even though I believe I selected all the same options.

On multiple occasions, I've done new NT server installs, that froze at some point during the install. When I repeated the install, with the same choices, the install then worked. Other NT experiences are documented at length in the Building Geodsoft.com section of the web site. My reading and talks with others, suggests my NT experiences are typical, rather than exceptional.

Obviously, I do not believe that either Windows NT or Windows 2000 is a highly stable platform. Before continuing with additional evidence to support, this I do wish to acknowledge that under the right circumstances, Windows NT or 2000 Server installations, can be made quite stable. There are many totally Windows based web sites, that are 7 by 24, mission critical operations for their companies. I believe that the Dell site is 100% Windows based, and this is an e-commerce and product research site, that is huge by almost any measure including transactions, dollars, and profit. You can't read the trade press or browse the technical web sites, without encountering references to companies that either entirely Windows based, or that have web sites that are entirely Windows based. If you claim that Windows servers can't be made highly reliable, or run major e-commerce sites, you are simply wrong. These sites typically have numerous redundant servers, and wrap Windows servers in balancing and failover technology. No one, except internal staff, will know how reliable individual servers are.

These sites have been developed by companies with large, highly trained IT staffs, that follow accepted procedures, and research the options and alternatives. As long as their sites remain up and their companies successful, no one from the outside can reasonably second guess their decisions, and say they should have done things some other way, or they would have had an even better site, if it had been done with different technology. The flip side of this is, that just because the choices they made were right for them, doesn't mean they are right for you or me. Unless their environment, goals, and needs are essentially identical to yours, you can't use their experience as anything other than anecdotal.

One occasionally encounters a statement along the lines that Windows servers are stable, and if they are not, it's because those administering them, don't know how to set them up correctly. A person saying this may be arrogant and simply have had very good luck with their Windows NT and 2000 servers, or they may really know what they are doing. Since some have succeeded in making stable Windows servers, it's reasonable to assume there is truth in the statement. On the other hand, I'm not stupid and have 18 years of professional experience, on a wide variety of small and mid size computers. My first attempts to install Linux and OpenBSD servers resulted in almost totally stable machines, while I've spent over four years trying to figure out how to install a stable Window's NT server, to conclude that it is beyond the resources I am willing or able to devote to the task.

As this document was nearing final form, my Windows NT server self destructed after the application of Microsoft's most recent security patch. A SANS e-mail I received soon after I installed the patch, the post SP6a "rollup" security patch, said SANS had heard the rollup patch "crashed a ton of systems and created a fair amount of general chaos." Even though the NT server provided essential services on my network (backup via CDR), not provided by other systems, I decided it was time to migrate these to other machines, and stop sinking time into this server that's caused me more maintenance headaches than all my other machines combined. As of August 20, 2001, all my public web servers as well as any other service I may make available in the future, will be based on UNIX variants only.

Top of Page - Site Map

Copyright © 2000 - 2014 by George Shaffer. This material may be distributed only subject to the terms and conditions set forth in http://GeodSoft.com/terms.htm (or http://GeodSoft.com/cgi-bin/terms.pl). These terms are subject to change. Distribution is subject to the current terms, or at the choice of the distributor, those in an earlier, digitally signed electronic copy of http://GeodSoft.com/terms.htm (or cgi-bin/terms.pl) from the time of the distribution. Distribution of substantively modified versions of GeodSoft content is prohibited without the explicit written permission of George Shaffer. Distribution of the work or derivatives of the work, in whole or in part, for commercial purposes is prohibited unless prior written permission is obtained from George Shaffer. Distribution in accordance with these terms, for unrestricted and uncompensated public access, non profit, or internal company use is allowed.