RDBMS and Kevin Bacon
Testing a Relational Database Management System, (RDBMS), can be like playing the
6 Degrees of Kevin Bacon Game.
Pick nearly any actor (or actress or movie) and a relational database of movies,
actors and actresses and a computer can search the database and find a connection
between Kevin Bacon and that actor (or actress or movie) in less than six jumps.
A CNN Article from October of 1996
about this lists an example;
"....works with almost any actor, from the obscure (i.e. Shirley Temple, Orson Welles, Errol Flynn) to now
infamous icons of pop culture. Take O.J. Simpson, for example. "O.J. Simpson was in
'Naked Gun' with Priscilla Presley, who was in 'Ford Fairlaine' with Gilbert Gottfreid, who was in 'Beverly
Hills Cop II' with Paul Reiser, who was in 'Diner' with Kevin Bacon."
On the surface this seems quite simple...
But, look at the method and pattern of matching.
At each step of the way the computer must decide if it can follow a relationship.
With no indices and two keys, "movie title" and "actor/actress", the engine must execute numerous
searches in a very processor intensive test, build, retest, rebuild manner attempting to build the
relationships from query to Kevin Bacon.
But, let's introduce a few incides and test and reduce the processing time.
First, lets build a filmography of Kevin Bacon's works.
This will make it a single "jump" single result query.
Next, lets build an index of all the actors and actresses who appear with Kevin Bacon in a movie.
Again, we will have a single "jump" single result query.
Next we will need to look at our data and decide, based on the overall number of movies and actor/actress names,
what other indices we should create.
|
|
|
Copyright ©2003 RMH, all rights reserved.
|
