Severe airline reservations - use of connectivity-based databases - Soft Machine

Chief Executive, The, Sept, 1992 by Robert Bittlestone

Bob: Well, Kevin, I understand you have a $50 million information systems proposal for our board meeting today?

Kevin: Yes, Bob, I want to walk you through our new process re-engineering downsizing project. We call it "Managing Our Ongoing and New Systems: the Holistically Integrated Nanosecond Enterprise."

Olaf: Forgive me for asking, Kevin, but doesn't that spell MOONSHINE?

Kevin: That's just like you planners, you won't take anything from I.S. seriously.

Bob: OK, guys, let's keep it friendly. Why don't you start, Kevin?

Kevin: First I want to explain to you the advantages of real-time computing; then I'd like to tell you how we can achieve this at a fraction of the cost of the 1980s.

Olaf: But, Kevin, shouldn't we start by discussing whether we want real-time computing at all?

Kevin: Go back to Frisbees and Slinky Springs, Olaf. We have some business to do here today. I'd like you all to think for a moment about the way you book an airline ticket. Wherever you are in the world, you can walk into an airline office or make a local call, and find out exactly whether seat 33C is free on the flight you want. Not last week, or yesterday, or a minute ago, but now! Right up to the nanosecond. Have you ever found yourself sitting on another passenger's lap on a flight to Denver? I'll bet you haven't! And you know why? Because of Codd.

Bob: Because of whom?

Kevin: Ted Codd, father of the relation data base. Codd taught us that a piece of data shall only ever live in one place. His laws have been handed down from one CPU generation to another. Thou shalt not duplicate thy neighbor's bytes! Praise be to Codd!

Bob: Tell us about the proposal.

Kevin: Well, it's obvious you should never duplicate data. You'd have to be sick to want to duplicate data in two places, when these days you can store it in one place instead and transmit it internationally in real time. Now the cost breakthrough I want to tell you about is that we can use a network of 100 downsized minicomputers with PCs for the end users, at a fraction of the cost of our previous supercomputer complex. The global communications network will only take about three years to set in place, and then we'll be up and running.

Olaf: What about the Bible?

Kevin: Olaf, why won't you be serious?

Olaf: I'm dead serious. You just said that a piece of data shall only ever live in one place, right? Well, at the last count several hundred million copies of the Bible dotted the world, each guilty of the most flagrant case of data duplication.

Kevin: Olaf, use your brain. The Bible doesn't change--it's what we call "read only" in our I.S. group--so it's OK to duplicate that.

Olaf: Kevin, your computers have only been around for about 40 years, so your profession can be forgiven its ignorance. But entire nations have been won or lost on updates to the Bible. The Hebrew Old Testament; the Greek New Testament; the King James version; the New English Bible; why, it's being updated all the time.

Kevin: Sure, Olaf, I know that, but it's not changing often enough to justify an online source. Also, the readers don't modify the text. It's obviously simpler and cheaper to publish and duplicate the Bible than to distribute it globally online. Although, wait a minute. Why not? The Distributed Deity? Time sharing with Him? I see a heap of new possibilities here.

Olaf: Not to interrupt your train of thought, but I've recently been hearing about an alternative approach to online data bases. I'm no expert, but it sounded simple. Gunter, you were at the seminar; what was it called?

Gunter: Well, I'm only a humble production engineer, but I recall they were comparing two approaches: connectivity and replication. Connectivity is where centralized data bases are distributed via permanent online connection. Replication is where disconnected data bases are synchronized periodically through automated dial-up. It was explained that historically, because a small number of the big I.S. success stories you read about (the airline reservation systems and so on) are based on connectivity, programmers have come to think all systems should be built that way. But apparently the real world is more suited to the replication model. It turns out that in practice you hardly ever need to know what everyone else is doing instantaneously.

Bob: Sure, I'll go along with that. If you did, you could never delegate; you might as well do everyone's jobs for them.

Kevin: Look, guys, this is complete heresy. How can you be serious? Why, next you'll be telling me you don't need to know about the results of our global subsidiaries in real time.

Bob: Absolutely right, Kevin. Once a month is just fine. Say, I like the sound of this replication business. Is it expensive?

Gunter: I don't think so, Bob. In fact there are PC-based tools out there already that can be set up by end users.

Kevin: Over my dead body! Connectivity at any price! Say no to PC anarchy! Uncontrolled propagation of deadly viruses! Crazed bands of mutant end users stalking vampire-like through the information infrastructure! [foam appears at the mouth] Gigantic bat-like figures swarming like killer bees through the tape drives! Hordes of . . . [collapses on the floor, clutching notebook computer to his breast]. Mommy! The umbilical! I forgive you! [faints in embryonic posture].