How do we create secondary indexes in Accumulo?

Question

I was reading the Accumulo documentation. I came across secondary indexes. Are these separate tables we need to create or these are internally created?

Answer 1

Secondary indexes are not automatically created in Accumulo.

They are purely a concept which you, as an Accumulo user, are free to implement, depending on your application requirements.

Answer 2

There are a couple of common options here (neither is ideal):

1. create a 2nd copy of the entire table, but indexed on the secondary column(s)
2. create a 2nd table that is simply a key/value lookup where the key is the value of the secondary column(s) and the value is the rowid of the corresponding row in the primary table

I expect the 2nd option to be more palatable soon given that, to quote the Accumulo/Hive Integration documentation:

Starting with Hive 3, indexing has been added to Accumulo-backed Hive tables. Indexing works by using another Accumulo table to store the field value mapping to rowId of the data table. The index table is automatically populated on record insertion via Hive.