Issue while creating a Couchbase Data Model

I am planning to design a Couchbase Data model which would replace a legacy system that runs on RDBMS. All seems to be good except, I can’t really figure out whether I should consolidate/arrange similar pattern data in multiple buckets or in a single bucket by defining types for them. Although the single bucket with different types seems to be the Couch-conventional approach, but I don’t really find a disadvantage in having multiple buckets.

Could anyone suggest me whichever would be the best modelling strategy?

I didn’t find any related document stating the cons of having multiple buckets.