I've search a lot to find any explanation regarding the key distribution but can't find something clear.
If I've a key based on a <long>:<string>:<string>
the long represent a datetime and the other string just Ids. This is the best solution i found to represent a table with primary key based on 3 fields

In this case what should be my InitialToken to have a repartition based on the date.
If I need to know exactly where the data need to be stored and of course I don't use a full replication on each server how do I force cassandra to distribute the key like this.

I hope it's clear.

Thanks for help