Dataloader/MP Reference Manual

Creating a Customized Version of DataLoader/MP
DataLoader/MP Reference Manual—424148-003
5-27
Generating Test Data
Generating Test Data
DataLoader/MP is useful not only for loading data from an existing system, but also for
creating test data. DataLoader/MP has several utility procedures that help create test
data. There are three requirements for test data:
•
You must be able to generate data in quantities equal to the quantity of real data.
•
You must be able to control the distribution of data so that it will be representative
of real data, or at least what you believe real data will look like.
•
The data values themselves must have some resemblance to real data. For
example, a name field containing the following test data might provide the bulk and
maybe even the distribution needed, but does not resemble actual data:
AAAAAAAAAAAAAAAAAAAAAAAAA
AAAAAAAAAAAAAAAAAAAAAAAAB
AAAAAAAAAAAAAAAAAAAAAAAAC
...
ZZZZZZZZZZZZZZZZZZZZZZZZZ
Reports designed with this data would look very different when generated with real
data. A client tool looks and operates differently when used with nonrepresentative
data.
To use DataLoader/MP to create test data, replace the default GETNEXTRECORD
user exit with one that generates data, using some of the procedures below. This
GETNEXTRECORD will build a record consisting of fields that each have one type of
test data. Each of the fields of this test data is likely to be of one of the following types:
•
Constants
•
Random numbers
•
Odometers
•
Cartesian fields
Constants
Some data fields might need to have constant values. All programming languages
support constants, so DataLoader/MP does not provide utility procedures to support
constants.
Random Numbers
No computerized method exists to generate truly random numbers. What is generated
is more properly known as a pseudo-random number series, which is a series of
numbers that looks random but ultimately repeats and has a very obscure pattern.
Not only do pseudo-random numbers repeat patterns, but each one is predictable
given the previous one and the algorithm being used to generate them. If a high-quality
pseudo-random number generating mechanism is chosen, such pseudo-random
numbers are almost always acceptable and are preferable to true random numbers