How is estimated key calculated in scylla. How does compaction strategy or RF affect it?

Question

I can get the number of estimated row using api family/metrics/estimated_row_count I wonder how accurate is this number and how could it be affected by compaction strategy and RF?

Say I actually have N rows, so I guess with RF = 3, it will report roughly 3*N rows?

Answer 1

The number should be the unreplicated one, so it's not a function of the RF.