The Old Joel on Software Forum: Part 4 (of 5)

Data generation

I'm looking at a relatively short contract for doing some minor ERP integration work, and it's a bit out of my usual experience.

Is there an application for generating a whole pile of data that I can use for testing the database? There is about 15 tables, with standard customer/contact tracking, manufacturing, sales tracking, etc.

I'd like to just throw a few hundred customers in, with realistic phone numbers, etc., and have a little bit of customizability in regards to manufacturing and sales data.

Is there anything like that out there? I was sure there would be, but google isn't giving me much.

Thanks,

Nigel
Thursday, March 11, 2004

I did something like that (in Access mind you)

I created a First Name, Last Name Middle names tables.

I then filled them in with fictious names.

Then I used a cartesian join to produce every combination and appended to another table - for example it might have produced 10,000 combinations.

Then I created a simple function that randomly picked from the full sample data to produce a nice subset of test data that was inserted into the test system.

You could do the same with company names, addresses and phone numbers.

YMMV

DJ
Thursday, March 11, 2004

Sounds like a good start.

I was hoping there would be a tool that would also allow specific distributions for the data, such as sales.

For example, I could throw in sales data where quantities and dates were distributed on a slight exponential curve. Probably not necessary for testing, but it would look good on sample reports that I'd be throwing together for the Client.

Nigel
Thursday, March 11, 2004

Sounds like a market niche to me.

Kyralessa
Thursday, March 11, 2004

I've done something similar in the past, but it was much more specific. You could specify a poisson distribution, or whatever your heart desired. As well, each text field had a length specifier and length distribution, so that you could specify 80 characters max for a last name, with a mean of 20 chars, for example.

In this case I was (or am) just so sure there was (is) something out there like this already.

Nigel
Friday, March 12, 2004

You could take a look at http://www.upscene.com/ at their Advanced Data Generator.

Greg
Friday, March 12, 2004

Thanks Greg.

Nigel
Friday, March 12, 2004

http://www.interbaseworkbench.com/

Ron Porter
Friday, March 12, 2004

http://www.quest.com/datafactory/

Andrew Hurst
Friday, March 12, 2004