Reminiscing about First Pacific Networks
Posted 2020-09-08 19:46 PDT | Tags: hardware tales
In 1996 I was out of UCSC, living in San Jose, and needing a job.  My buddy Tash landed me a gig at First Pacific Networks as a junior system administrator, and that paid rent while I figured out how to get a programming job and kickstart my career.
FPN built these nifty little cigar box sized units which plugged into a broadband network, hijacked part of its band, and used it to stream voice, video and data (ethernet).  They also made headboards which plugged into a central office and managed the network.  It turned out that some countries (like Russia) there were extremely good broadband networks for squirting video around, but their telephone and data infrastructure were for crap.  FPN's project was an elegant solution that leveraged this existing infrastructure to provide the rest of it.
Well, at least it was "elegant" on paper.
The box had an RJ11 socket for plugging in a landline telephone, and that worked quite well.  It also had an AUI ethernet port, which worked rather less well.

The AUI port

My clearest memories of FPN was of those AUI ports.  Mostly that they were horribly, horribly unreliable, and hard to fasten/unfasten without bending the little clasps that flanked the socket, which of course made it even less secure.  They were always falling out, and our device drivers did not handle random disconnects well at all.
Part of the problem was that the AUI cable itself was really damn thick and stiff, so just bending it into a U-shape from the back of the box to the back of the PC put quite a bit of flexural tension on the whole rig.  Bumping the desk would often be enough to send the whole thing sproinging loose.
The temptation was to just glue the cursed things in place, but of course that was a no-no.  Such is the life of a junior sysadmin.  That, and replacing dead monitors, fixing power supplies, and rescuing borderline-tech-illiterate employees who deleted COMMAND.EXE or CONFIG.SYS to "save room" on their PCs.
Fortunately I didn't have to suffer that long.  I automated away some of my more annoying tasks when I could by writing software, and after a few months of that I was able to transfer to FPN's engineering team as a junior software engineer.
It was the break I needed.  After about half a year of writing network protocol extensions and simulations to demonstrate the scalability of our network (or lack thereof), I accepted a job at Cygnus Solutions as a GNU toolchain engineer, fixing bugs in GCC, GAS, GDB and the like.  My career had finally begun.