Iodynamics just finished a very substantial job for a client. We migrated them from an ancient SCO Unix v3.2 system connected to dumb terminals via serial lines to a Linux system connected to state of the art workstations via 100Mbps ethernet.

It was a remarkable task and we did a remarkable job of tackling the task. In the wrong hands, this project could have resulted in an expensive loss of critical data for the client or could have been marred with delays and exorbitant costs.

But, it wasn’t. Everything went smoothly and within 10 days of contacting Iodynamics, our client had a new Linux server with their business application loaded with all their current data ready to go. They also had 15 ethernet jacks professionally installed in the walls throughout their offices, a 24-port patch panel, and a 24-port 100Mbps switch.

I was thinking about this project as I drove home from the client’s offices this afternoon and it occured to me that so much of my past work experience was crucial to the success of this project. This was one time where jobs I’ve had in the past really paid off.

The SCO system had no network connectivity. At the time the system was built, networking was an additional, expensive add-on. The SCSI subsystem was exhibiting strange problems that prevented us from using the built-in tape backup drive or another backup drive. The only removable storage option was a 3.5-inch floppy drive. As a result of all these circumstances, getting data off this system was not going fun. My last option was to use one of the serial ports.

Now, I’ve dealt with a few serial devices in my time. When I was first in school at Utah State University, any device—be it a PC or a terminal— attached to the VAX system used for e-mail, was connected to a serial cable that usually went to a multiport serial multiplexor.

Because I was working for Dr. Joe Doupnik, the author of MS-DOS Kermit, as a computer lab consultant, I was able to learn first hand about using the Kermit file transfer protocol for transfering data across serial lines. I also learned the ins and outs of using MS-DOS Kermit and C-Kermit as terminal emulators for serial communications.

In 1992-93, I was employed as a system administrator for a company that produced automotive accessories. Their system included a multiprocessor Unix SVR4 system connected to a series of Emulex terminal servers which served dozens of PCs, Wyse dumb terminals, and modems. I learned all about the command-line and telnet interfaces on the Emulex devices for configuring the serial ports in the Emulex servers.

I also learned how to assemble, solder, and repair serial port connectors.

In 1994, I was employed by a good sized electronics manufacturing company in the Salt Lake City area as a database programmer and a system administrator. This was an IBM AIX shop running a centralized manufacturing business database software package based on the Progress database system. The users were attached to special multi-port serial devices made specifically for the RS6000 hardware.

While at that company, I began their transition away from serial toward ethernet. In the case of printers, this involved installing ethernet print servers. I became a fan of the Lantronix line of print servers because they included a command-line and telnet interface much like the Emulex terminal servers I had used before.

Since that time, I’ve learned a lot about Linux. There are still plenty of people who know a lot more than me, but I’d say I know quite a bit.

So, back to this project.

To get the data off this SCO box, I had to use the serial ports connected to a Digiboard multiport serial device. The ports were configured to run at 9600 bps and 19200 bps. The data we had to get off the SCO system was about 400MB. That’s not a lot of data until you consider moving it over a 19200 bps serial line. Then you’re looking at 2 days or more.

When I had exhausted all other preferred options, I logged into the SCO system and reconfigured one of the serial ports to its maximum speed: 38400 bps. Then, with luck, I found a rudimentary kermit program on the SCO system. I used tar and compress to created compressed archives of the data I needed to transfer. The resulting archives totalled 130MB.

Using C-Kermit on my laptop, I was able to transfer all the data in a matter of a few hours (I left it running overnight). Then, I transfered the data to the Linux server (via ethernet, thank goodness) and extracted the data from the compressed archive files.

There were two printers attached to the SCO system via serial lines. To keep them working, we got a Lantronix EPS2-100 print server with two parallel ports and two serial ports. It was just like old times logging into this device and setting things up.

To attach the serial printers to the Lantronix print server, I had to create some cables that had modular RJ45 connectors on one side and RS232 DB25 connectors on the other. This was nothing I didn’t know how to do with a soldering iron. Both printers worked fine with the first test print job sent to them.

This morning, everything was ready for the software consultant to log in (via SSH over a dialup PPP connection I set up) and configure the business application and load the data. After he had been working for about ten minutes, he called me (he’s on the east coast) and told me he was having a great day; everything was going just fantastic and smooth.

There were little hiccups with printer subsystem commands and terminal emulation software on the workstations throughout the day, but, for the most part, the users were able to get their work done without any kind of inconvenience at all.

I think today is one of the few times I feel really proud of what I’ve done with my life so far. I feel privileged to have gotten the opportunity to use so many skills I’ve picked up over the years all on one project.