Perry's Checkered Career

• Hewlett-Packard, Computer Support Division, 1980.
  Worked on remote support for the HP1000. However, the product division cancelled the HP1000 project, which cancelled my project. In one of life's ironic twists, CSD later becames the Response Center Lab, which then spun off the Customer Support Lab (see 1992).

• Hewlett-Packard, Information Networks Division, 1981-1984.
  Worked on data communication products which linked the HP1000 to IBM mainframes
  • RJE/1000 (sustaining only), an emulator for the venerable IBM 2780 Remote Job Entry station, which consisted of a printer, card(!) reader, and card punch (!!).
  • MRJE/1000 (drivers). Seven RJE stations on a single datacomm line. When it works, it looks really cool on a protocol analyzer.
  • RJE/1000-II (Principal architect, drivers, Z80 firmware). A re-write of the previous product, which eliminated an unbuffered serial interface card that had real-time I/O problems.
  • PMF/1000 (consultant to the consultants). PMF is an IBM terminal emulator, which allowed interactive (non-batch) datacomm.

  ... meanwhile, IBM was developing SNA, a saner peer-to-peer protocol.

• Hewlett-Packard, Ft Collins System Division, 1985-1991.
  • Kernel serial datacom drivers. Ported some of the McKusick file system.
  • Spiderman (a hypertext help tool that died in 1988 for want of NCSA Mosaic and HTML)
  • SAM (a.k.a. System Administration Manager) kernel config and the first s700 SAM.

• Hewlett-Packard, Workstation Systems Group, 1991-1992.
  • Worked in Technical Marketing's customer problem escalations group, resolving kernel problems. Escalations are customer problem with an HP product that are affecting the customer's bottom line. This job is one-third technical, one-third diplomatic, and one-sixth Dear Abby, and one-sixth keeping a cool head and your sense of humor.
  • Worked in my spare time (NOT!) as Mr Patch, who administers the HP-UX patch server - all 30 gigabytes of it.

  This group got nuked when Expert Centers came into vogue. Consistent with The Law of Irreversible Reorganizations, Tech Marketing did not reappear when Expert Centers were reorganized (See 2001).

• Hewlett-Packard, WRCO Customer Support Lab, 1992-2001.
  • Single-handedly supported the X server drivers as well 10 million lines of code in X, Starbase, Windex, and any other graphics that needed a bug fixed.
  • Worked on ProPatch, a proactive patch recommendation tool used by HP's Response Centers. ProPatch was originally developed by the WRCO Mayfield Response Center Lab.
  • Transitioned and supported COP (Customer Operational Profile), a high-tech webbed Rolodex for HP's high-end support contracts that was implemented in cgi-bin/Perl. ProPatch was eventually mind-melded into COP.
  • Created an automated firmware analyzer and analyzer database for the High-Availability Observatory, which was supposed to replace COP (it didn't quite, requiring support of BOTH).
  • Supported four patch tools (roughly organized by HP-internal/external and proactive/reactive.) Moved these tools toward the Grand Unified Theory of patching, since underneath it all, there are only so many ways to walk a chain of superseding patches. Also fixed the patch distribution mechanism, which wasn't reliably sending patches to the HP Support network.

• Hewlett-Packard, Worldwide Technology Expert Center, 2001.
  A reprise of my 1991 Tech Support job, except on steroids.
  • The Expert Center resolves any customer calls that the Response Center front lines cannot. If the Expert Center can't figure it out, the problem goes to the Lab Engineer who wrote the code. I did kernel calls. HP-UX has a BIG kernel.
  • While I was taking calls, I also wrote a web-based e-service that correlates data on four databases and identifies products with support problems. Management can then allocate proactive resources to correct the problem.

• Software product development. Know your customer before you code it.
• Customer satisfaction. Know your customer before you ship it.
• Drivers. EEs understand bits flying across a bus, even if it's done in software.
• Software defect removal. Finding the needle sometimes requires burning a lot of hay.