Generation of Data with Zipfian Distribution for Database Testing

It is often very hard to perform testing on real data in large databases. In order to execute testing of some performances on databases, data need to be generated synthetically by using distribution that suits to the real data the most. In many papers authors state that the data has been generated by using Zipfian distribution with appropriate z values, but they don’t explain the process of generating such set of data, they just give pieces of complete scenario. We have collected those pieces and put some effort in comparison of results they have published with ours. In this paper we present the process of generating the synthetically data for testing of accuracy of approximate queries in large databases based on knowledge we have gathered from papers and we have discovered by ourselves.