Need info on how to handle 1 billion keys

[kharabela]

My use case if to reference person's first name, last name & SSN to uniquely identify .Insert that record in key value on if thats unique.If my records grow to 1 billion I want to search unique person and insert only if not existing

[derekcollison]

Currently subject based indexes for lookup are held in memory.

So I would suggest using a hashing approach to use multiple streams / KVs and size accordingly with memory.

We are looking at improving the memory footprint for very large key/subject spaces.