Difference between recored inside set and without set

Operations Planning
1

Aerospike supports record within set and directly within namespace (without any set).

What are the use cases for the later one?

Is there any performance or functionality limitation on them?

My case: I will have lots (1 Billion) of records in a set say test_set within a bin named test_bin_value. Now I want to keep track of latest record update/insert time in bin update_timestamp. This bin’s value will be updated whenever someone insert or update record in bin test_bin_value. Should i keep bin update_timestamp inside set test_set or outside? If i will keep it within same set, will it take more space? There will be frequent read and write operation on both bins.

2

These questions are answered here: Capacity Planning Guide | Aerospike Documentation

Originally there was only the null set.

From the link above

If using a set name: [for each record add] 9 bytes overhead + set name length.