We're trying to find an acceptable way to clean up production SSNs for use outside production. I've been looking at one way hashing with salt. I've looked at the HashBytes and Checksum functions. I can use these to generate a pretty good 4 byte hash of an SSN.
I'd really like to end up with a 9 digit number so the application actually works. Has anyone done anything like this? Any thoughts on how to accomplish this? It would need to be compliant with all the various privacy standards.
My data set is 100 million rows (20 million unique SSNs) and the SSN appears in multiple tables (yes, I know that's an issue). I'm starting to think that I may just need to build a temporary lookup table and generate 20 million or so random SSNs and then do updates to the original table. After that I can drop my original mapping table.