Implementing a hash key collision strategy


Dan Linstedt, the inventor of Data Vault, has written a lot about it: hashkeys.
For instance, one of his latest blog posts:
#datavault 2.0, Hashes, one more time.

I will not list all other sources, as you can use Google yourself.
A few comments on hash keys:

  1. You need them for scalability. Using sequence numbers is taking the risk that your data warehouse does not scale well later when the amount of data grows.
  2. They can collide: two different business keys can produce the same hash key. However the chance that this happens is very small. For instance when using SHA-1 (which produces a hash value of 160 bits) you will have a 1 in 1018 chance on a hash collision when having 1.71 * 1015 hash values (read: hub rows) according to this blog post.
  3. If collisions are unacceptable you need a hash key collision strategy.

The full article is posted on DWA.Guide, so you read further there ..

Picture credits: © Can Stock Photo / alexskp