In my current project I have to carry out a two stage replication (this is a strict requirement). I'm looking for the recommended approach to this. Stage 1: Extract from MySql --> S3 as Parquet files Stage 2: Extract Parquet --> Load to Redshift for transform

I've never tried this, but I think you are able to chain the extractors and loaders i.e.

tap-mssql target-s3-parquet tap-s3-parquet target-redshift

if not, then you'll have two jobs to run

tap-mssql target-s3-parquet
tap-s3-parquet target-redshift

🤷