octavia shock I am trying to understand why my table has a different number of rows in the destination. I am doing this experiment with only one table which has 7.5m rows in the source. Replicate mode is

increment|deduped history

and in my destination I have 3 tables (normalization enabled): • raw_table: 2.5m rows • table 2.1m rows • table_scd 2.6m rows I expect that scd and raw table should have the same number of rows. the main table will have the deduplicated number of rows. What am I missing on this?

raw tables append all records (new ones and old ones) table_scd stores the history of all records (new ones, old and deleted ones too) table the final table with dedup record

@Marcos Marx (Airbyte) another question, in the case of using

increment|deduped history

but without normalisation. how the resulted tables will show data? would it be the same structure: raw tables append all records (new ones and old ones) table_scd stores the history of all records (new ones, old and deleted ones too) table the final table with dedup record except the rows will be having josn blobs instead of normalized rows and columns?

you must use normalization to achieve increment + dedup history, there isn’t an option to use increment without dbt/normalization

You are not using

cdc

, so how come table_scd will save deleted rows ? As per document,

incremental

mode does not support deletion detection