Destination Postgres: normalization is slow (CPU-bound, using 1 core)

[ggam]

Tell us about the problem you're trying to solve

I'm replicating some Oracle tables to Postgres destination. One of them has 160 columns and normalization becomes a bottleneck as Postgres only uses one core to process it.

Normalizing 930k records took more that 24 minutes:
2023-04-23 14:38:06 �[42mnormalization�[0m > 16 of 16 OK created table model xxxx.xxxx................................................................. [�[32mSELECT 930527�[0m in 1445.07s]

Describe the solution you’d like

Postgres won't parallelize the query as it's a full table select, so I guess the easiest solution would be to:

• Convert the raw source table into multiple staging tables, so dbt can execute concurrently.
• Combine the temporary tables into the final one.
• Drop the staging tables

Describe the alternative you’ve considered or used

If and when #25194 gets done, that should solve the root problem.

Additional context

I can privately send the workspace dir and more information about the tables if needed.

Are you willing to submit a PR?

I don't realistically think I could offer my help on this one.

[evantahler]

Closing this issue because normalization as we know it is going away shortly - learn more @ #26028