I am using a very large dataset and this technique for generating a unique id for each observation doesn't work because _n caps out somewhere in the +e8 range. Is there a quick and easy alternative or a way to correct this?

Thanks.