Rationale for choices in the Diskless Workstation Project

Although I said in the introduction that some of the motivation was to "do it my way." I have always been one of those kids who takes the toys apart to see how they're made or how they work. This includes computers. I have been involved with hardware design since about the time that the 7400 came out. My involvement with software even predates that by a couple of years. Through various jobs, I have gained a basic understanding of operating systems even writing a couple of real-time OS systems that ran many years before being retired. This has involved interfaces of various types - serial, parallel, analog, etc. However, these rudimentary OSes never had to deal with any mass storage or real communications. ('Real' meaning anything that would come close to being networking.) And now that computers have become commodity items, I finally have a 'toy' where I can learn how the innards work.

So when I tackle a task like this I tend to not take the easy route. I want to get may hands dirty and understand what's under the hood. So far, I've tried Slackware, SuSE, Caldera and Red Hat, in that order. I've even deployed several systems utilizing Red Hat related to my work. However, I tend to return to Slackware when I want to understand how the system works. I acknowledge that there are some disadvantages to this attitude, but I see the front ends of most distributions as facades that hide what I want to see and find them just frustrating. Don't take me wrong on this. The GUI has its place and it's certainly needed for Linus to make inroads to a general consumer market. It's just that I'm not an average consumer. So, to start off, I've chosen to work with Slackware. (Version 7.0 to be specific, just because that's what I had on hand when I started this project.)

One of the other items that I had to choose an approach on was whether to use the BOOTP or DHCP protocols. I decided to go with BOOTP. Partly because it was simpler. The documentation seemed to be aimed more for at this type of application. As a result it seemed to be clearer. Also, I'll quickly admit that this was the quicker (read lazier) approach at the time.

Finally, there was my choice of hardware. The majority of it is remnants of upgrades required by the 'other' OS. One of the machines started life as a '386 for my second son as he went into college. The other was a '486/DX4 that started life as a system for my wife to use for her business. The '386 system has lost it's hard drive and thus was a natural for the diskless system. The '486 system has been expanded in memory and disk space as well as adding a CD-ROM to help in the installation process. As I've needed hardware, I have been able to get what I needed on the used market. Mostly from DeepSpace Technologies here in Frederick, MD. The Ethernet cards are ISA cards that were removed from service at work when the move to PCI cards was made several years ago.

Copyright © 2001 David J. Pfaltzgraff, all rights reserved.