Just recently I realized that my local computer installations had gotten somewhat complex. I have a long background working with different operating systems (OS) and as a result, I don't have much difficulty working with different systems. Yesterday I realized that even my usual multi OS environment had gotten complex.
Let's make that very complex. As I write, I have two OS/2 systems, two FreeBSD 4.9 systems, one Windows NT, and one Suse 9.0 online on my local 100 Mbs network. In addition to that complex, I have an idle system that will run FreeBSD 5.2, another set of spare parts that will run RedHat 8.0 and OpenFiler in an experiment, and a planned OpenBSD 3.4 that will be a firewall.
Six of these systems are attached to two switches driving two monitors, keyboards and mice, which constitutes my desktop. The others are on the workbench attached to a five port ethernet switch, which is attached to an eight port ethernet switch at the desk, which is attached to DSL. While this is a fair number of systems, a friend of mine has double that number, but only three or four operating systems.
The best description of how it got this way is "It just growed." I didn't set out to use six different OS. It all started in 1981 with my first personal computer. No, not the classic IBM PC, but a 16KB Radio Shack Model 3 with a casette tape. For a mere $1,000, I could start at the bottom.
I started with a Radio Shack Model 3 in 1981, followd by a Sanyo 550, a semi clone of the IBM PC, followed by a Sanyo 880, a real clone running at 8 MHz. I ran DOS for those systems, with a later AT clone running Dos and Desqview, still later all versions of MS Windows from 1.01 to 3.11, upgrading to a 386/25 in 1987. I continued to use Windows, despite Desqview being a better OS for multitasking. Microsoft was deliberately making incompatible changes to break Desqview with every release. Back in those days, fixes took days or weeks to get with a floppy in the mail. That was my earliest experience with MS trickery and monopolistic practices. As of 2003, they have not changed, simply become more subtle.
I finally standardized on OS/2 2.0 after testing the beta in 1991. At that point I was running on a SOA 486/25 with 16 MB - the base system cost $6,000 with 8MB and a 70MB disk. I think I still have that motherboard stored somewhere - it's hard to part with that much money, even an obsolete MB. The computer tower case it came in has been in continuous service since 1992 and is currently housing the Suse 9.0 system. The cheap stuff you buy today would have failed in less than five years.
After four years of running the 486 system, having added a $2700 680 MB hard drive in 1993, I worried about backup if my only system went down. The Pentium clones were just out then, so in 1996 I bought some parts and built a cheap backup system, keeping my primary system the 486. During that time I had bought a speedy 1200 baud modem for a mere $500, a single speed SCSI CD for $315, including controller and cable, and a 1 GB SCSI hard disk for about $300. I thought I would never run out of space, or for a couple of years anyway.
A few words about why I have continued to use OS/2. One reson why the 486 lasted so long was that OS/2 enabled very effective multitasking, so that I could run long jobs in the background and still get good response on my interactive job. Another reason was that the system simply did not freeze or crash, despite heavy use with multiple programs running - downloads, computation, printing and interactive simultaneously.
To test this reliability, I ran a load test by starting a 200 page Pagemaker print job in the Windows 3.1 subsystem, a streaming tape backup on a SCSI tape drive, the Windows Solitare game in auto play, and an online modem BIX session as an interactive task. The tape backup ran without pausing, the BIX session was responsive, the LJ 4 printer worked fine, and I could see the Solitare game running on part of the screen. Microsoft didn't manage to match that until Windows NT, which used the OS/2 kernel codeveloped with IBM before the breakup.
When did the first OS in addition to OS/2 arrive? In the late 1990s, I bought a copy of Caldera Linux 1.1, and later 1.2. By that time I had more than four computers, not all of which were operational at the same time. I was switching components around almost weekly, except for the primary OS/2 machine, then a speedy Cyrix 166. My success with mobile parts was spotty, earning me the description of having a 'poisoned' screwdriver. That made as much sense as anything else.
My upgrade procedure was responsible for my component switching. When I wanted or needed an upgrade, I would buy something new for my primary system, usually replacing an existing component. Since my primary system always had the best performaing parts, I would move the replaced part to the second best system, and that part to the third best, and so on. A single part could involve me in a few days worth of upgrades for several systems, and it worked most of the time, but was extremely labor intensive. Eventually I would collect enough spare parts for another system, and buy a new or used chassis to put them in. In retrospect, I'm glad I didn't get into this with cars.
By 1999 I was running multiple OS/2 and Linux systems, with one Linux as a firewall. Around then I discovered FreeBSD, and the other BSD variants, including OpenBSD. I replaced the Linux firewall with OpenBSD 2.8, then tried FreeBSD 4.0 as a workstation. This was in response to IBM's announcement of End of Life for OS/2, which turned out to be somewhat premature. I was not satisfied with Linux as a replacement and liked what FreeBSD delivered, so I started my current setup using OS/2 as primary workstation, and FreeBSD as secondary. By 2000 it was clear that FreeBSD was quite complimetry to OS/2 in that it was reliable, multitasked well and gave me access to the growing armada of Open Source programs.
My long term plan was to gradually move over to FreeBSD as was comfortable, no rush involved. Even if OS/2 was EoL, it was still running reliably and would be getting fixes and drivers for a while. Then eComStation was announced, and OS/2 was reborn with independent support and IBM's cooperation and licensing. I didn't have to convert!
By this time I was comfortable with FreeBSD and liked the additional capabilities it gave me, so I continued with the two platform system, and OpenBSD as firewall. Three operating systems total. However, I was seduced by an announcement by then original SCO that the first 75 sign ups for SCO developer would get free OS code and licenses. I'm stuck with a very strong sense of curiosity, and I wanted to see what they were like, so I applied. I did get the disks, but due to changes in plan, never put them into a system. They still sit on the shelf, but given the current SCO attack on Linux, will never be used except as an experiment. I'm not about to support a company who does what the revised SCO is trying.
RedHat and SuSE are recent 2003 additions. During my daily wandering around the net in search of interesting and useful items, I found a new open source release of a storage management software named OpenFiler. While it was labeled beta, it was an release of a working set of code from a British company. I am currently trying to get a large storage server set up, and was interested in this for future developments. I downloaded the software, read the install which said it would currently only run on RedHat 8.0 with a strict install process. I downloaded RH8, wrote the CD and installed it.
I had to put that process on hold as I realized that I was not about to run a production server on beta software, even if it was from a working system. Time to rethink. I could run RH8, wait for FreeBSD 5.2, planned for December release, or what? RH8 was in hand, installed easily, but I had never been comfortable with RedHat and their recent dumping of all but paid enterprise systems, despite Fedora, left me cold. So now what?
Despite running FreeBSD, I was also watching the Linux distros mature - keeping an eye on the competition so to speak. I found SuSE (awkward capitalization) to be consistently focused and organized. I was a little concerned about adding Yet Another OS (YAOS) to my stable, but that was only a minor glitch with my pre-pc background (Story another time). I have now installed a mostly complete SuSE on my storage system, which was an 8 hour job over the 60KB DSL link. A bit slow that.
I'm impressed with my first view of SuSE 9.0. Unless I find a show stopper in the next few days, that will be what runs my big (90 GB Raid 5) storage system. I know 90 GB isn't much these days, but these are heavy duty, five year warranty SCSI drives, not the cheap junk sold as ATA drives. I really am a bug on reliability. Currently I'm creating a local mirror of the 9.0 distribution so I can install over 100Mbs local net. Total space required is about 7.5 GB, so I'll dedicate a 9GB drive to it. SCSI of course, courtesy of a friend's donation.
That brings this part of the story current, though there are numerous learning experiences left out for lack of time to write. I'll add items as I can, hopefully being both interesting and useful. Reader feedback is always welcome at bwrite $at$ ywave.com.