Phone: 952-947-0438

Resume Addendum II:

WorldFlight System Overview/Technical Background

Most of my experience at Northwest Airlines was obtained as a member of the technical team (approximately 35 people at the start in 1988, but reduced after the economic downturn in 1992 to a group of roughly 12 to 14 people) that did all of the development and support programming work related to the "WorldFlight" system, a large Unisys 2200-series mainframe-based transaction system which was the central or "core" computer system used by both the flight operations and the ground operations personnel at NWA, as well as the people in the SOC (shorthand for "System Operations Control", the operational nerve center for the airline).

As a member of the WorldFlight team, I was almost continuously engaged in the following types of activities:

The WorldFlight system was arguably one of the most "critical" systems at NWA. It was based on older Unisys technology (the original system, called UNIMATIC, was developed at United Airlines starting around 1967 on a UNIVAC 1108 and obtained by NWA in 1988), but the system was tremendously stable given the large and varied amount of activity that occurred on it from day to day. On the few occasions when the system was down, the airline took a lot of flight delays and ended up losing a lot of money. Even individual application glitches could result in flight cancellations or delays due to a lack of critical information. One of the few system outages I remember (caused when a fiber line was accidentally cut a few years ago, isolating the SOC in Building F from the WorldFlight system in Building J) even made the papers here in the Twin Cities because of the outage's impact on the airline's daily operations. Quite frankly, I felt that it was quite an honor to be one of the programmers supporting such an important system.

WorldFlight not only provided a large and varied set of interactive text-based applications that were used by many different employee groups at the airline, but it also automatically processed a very large number of message-driven "datafeeds" coming into the system from various sources (some inside the airline and some not) including weather alerts and other weather bulletins from the National Weather Service, passenger and bag count messages from the WorldSpan (PARS) reservations system, and both freight and cargo and container information from the Cardinal (USAS*CGO) system.

The application code resident on the WorldFlight system also performed a number of other tasks, including:

As you can see, WorldFlight did a lot of different things in a variety of different ways, and many of the applications resident on the WorldFlight system were important to the operation. The critical nature of the system was made a little more challenging by the "hidden" or automatic nature of much of its processing -- most of the processing of incoming and outgoing datafeeds was done completely without human intervention, and tracing problems in such an automated system was quite challenging, particularly when there were several different systems passing the data back and forth in real-time, and when WorldFlight environment had no real "debugger" as such at all, just traces.

The software technology used in WorldFlight was old, at least on the application layer, but the hardware it ran on was not, and the environment ran on top of OS2200, a mature operating system native to the Unisys 2200-series mainframe line. Most of the 2000 or so programs and 1500 subroutines that composed WorldFlight (roughly two million lines of code in total) were written in an older variant of Fortran, with some being written as newer Fortran modules or as ASM (assembly) routines. Most individual operations on the system were fast -- a transaction (generally defined as a user keyin at a terminal followed by a resulting display screen) was generally expected to take a half-second or less to complete, and some of those keyins were able to perform several dozen I/O's while doing so. It was very fast.

Software was created in close collaboration with the user representatives in the SOC. Most of the folks I worked with when doing either design work or testing were flight dispatchers, pilots, ground ops people, or folks related to NWA's meteorology department. I also did some work with other groups from time to time as projects required.

During my two years as a Unisys contractor with the WorldFlight group and my eight years as an NWA employee, I worked on perhaps 30 major projects and 100 or so smaller ones. The sheer variety which is present in this type of application programming work is very hard to summarize in a few pages of text, or in a resume, and in a system where error-free operation is absolutely critical, almost everything one does ends up becoming very important.

Most of the work that I did while at NWA was in the following areas of the WorldFlight system:

Some of the projects I worked on involved the design of new databases and program flows and the creation of completely new code, others required the modification or complete rewriting/replacement of existing code. Some projects were quite large by our standards (the Flex Thrust project was perhaps 5000 hours in total, of which I accounted for roughly half, and the SYSERR database took perhaps 1800 hours to complete as a solo project).

I also spent some time (roughly a year) supporting a set of daemons written in C and running under Solaris that accepted various bits of flight and gross weights data from the WorldFlight system and stuffed it into a Sybase database using CT-Lib functions. That gave me some additional exposure to C in a Unix environment (added to the playing I'd done at home under Linux with DDD).

Instead of trying to describe the routine work in my resume, I've decided to highlight only the most significant accomplishments, and to show only those projects which I felt were exceptional in some manner. Believe me, there's a lot more!

I hope this document provides some useful background into the nature and scope of the applications on which I've worked. If you have any additional questions, please feel free to contact me.

[Valid HTML 4.01!] [100% hand-coded HTML]

Most recent revision: August 13, 2003
Copyright © 2002, 2003 by Richard C. Steiner

Site Meter