Building The First-Ever OS Install CD

In the spring of 1993, while working at IBM Research in Yorktown Heights, NY, I managed a group that developed and supported an OS/2 application platform. Before work one morning while watching the Today Show, I learned that Philips was about to bring a consumer CD burner to market, with a list price of $6,000. At this point, the only way to burn a CD was by using a Sony solution that cost over $25,000. After justifying the expense to my management that morning, the hunt was on for a vendor willing to sell me the product, remember this was before the Internet and Google. I located a Philips Value Added Reseller in Atlanta who would provide not only the burner, but also the DOS software, and blank CDs. Now at that time, blank CDROMs were $45 each, they used real gold, and this price required that you buy at least 100 at a time. We're talking just over $10K all in to produce a CD.

The CD Mastering software was classic buggy version 1.0 code running on DOS, and it would often take me three or four tries to successfully complete burning a single CD. It was far more art than science at that point. My initial objective was to build a CD using a menuing system the OS/2 Tools team had developed for installing software over the network. These CDs would then enable us to provide installation from CD for other sites that wanted to use Watson's OS/2 CORE Platform.

Ten years earlier, while in college, I'd experimented with bootloaders on 5.25" floppies and had looked very carefully at how TRSDOS, LDOS, NewDOS, and several other OSes had worked. I then used this experience to explore how OS/2 loaded from 3.5" diskettes. The BIOS on IBM's PS/2 line of computers at that time didn't support booting from CD, so a single 3.5" diskette was required. I crafted an OS/2 lite pair of boot diskettes that would load enough drivers from the diskettes so that the hard drive, video, and CDROM devices could be recognized and used. At this point, I enlisted another member of the team, Damien Scott, who provided the final piece of code to trick OS/2 into thinking that the CDROM was the second diskette drive, and from that point on, it was trivial to finish the installation. This process enabled someone to receive a single OS/2 CD, and from this disc, they could then create the necessary two boot diskettes and install OS/2. Before this process, OS/2 required 12 diskettes to install, and that was just the base OS. 

We then worked with Tom Rogers to adapt his menu application for installing programs from a network drive so it would install them from a CD. Tom's program would enable us to distribute a single CD with OS/2 and all the programs someone might ever want; researchers could then install systems more easily regardless of where they were. All this progress caught the attention of IBM's OS/2 group in Boca Raton, Florida.

John Soyring, then IBM's VP of Personal Software Products, requested that I technically lead a team in Boca to build something he envisioned as the OS/2 Professional Developers Kit. So my wingman Dameon and I packed our bags for several weeks and delivered the first-ever OS/2 Professional Develop's Kit (PDK). Dameon knew the menu system better than almost anyone except its author, Tom Rogers. I handled content aggregation, documentation, packaging, mastering, and duplication. Dameon handled menus, miscellaneous scripting, and testing. The VAR in Atlanta had a CD press and could turn out 100s of copies an hour within 6-24 hours of receiving a master CDR from me. John Soyring had a busy show schedule that fall, and on several occasions, we were shipping out CDRs on commercial flights as cargo via Delta Dash. Soyring would receive a product level CD, and volume copies, often thousands of CDs, only several days after a master was burned in Boca and shipped to Atlanta for duplication. At one point early in the process, the OS/2 PDK CD contained 43 diskettes of code. All of IBM's programming languages, OS/2 tools, networking tools, editors, a version of MS Windows that could co-exist with OS/2. It was IBM's equivalent of the OS/2 kitchen sink.

News of the OS/2 PDK first leaked out in a Spencer the Katt's column in PCWeek. Then a few issues later we made the cover above the fold, which was still a big deal back in 1993. Here it was memorialized on the right with signed copies for everyone on the team. One would think that figuring out how to install an operating system from CD, something that had never done before, and for creating the first-ever OS/2 Professional Developer's Kit would have been worthy of an IBM Outstanding Technical Achievement (OTA) award. I had to wait five more years and switch divisions before receiving an OTA. After the dust settled, and a little pressure from IBM Watson Research, Boca's OS/2 group, came through with PC Company Divisional Awards for Damien, Tom, and myself for our work with the OS/2 team and production of the PDK.