Glue to automatically create target schema

Question

Can someone please clarify if GLUE is supposed to be able to automatically create tables in my target database before loading in the data?

I haven't been able to find anything on the options:
 - Create tables in target
 - Use tables in the data catalog and update your data target

when setting up a job to ETL from say RDS into Redshift.

Since the data has already been catalogued, it would seem the schema should be automatically created but I haven't been successful getting that to work.

Thanks!

Accepted Answer

- Glue does automatically creates the table schema is it doesn't exist.
However, any "string" type will be translated into a `varchar(65535)`, which is not efficient for a columnar database.
- If the table already exists, Glue will just add the records into the table.
- If Glue can't map a field into the table (example: you have a field "createdate" as string type in your python code and the table has a "createdate" as timestamp type in Redshift), Glue will automatically add a field "createdate_string" to the table and populate that field.

When Glue calls `glueContext.write_dynamic_frame.from_jdbc_conf()`, it stores the output dataset into the S3 bucket given by: `redshift_tmp_dir = "s3://jm-bank/tmp/"`. Then Glue performs a COPY command to Redshift, which you can visualize in "Queries" option from your Redshift cluster.

Glue to automatically create target schema

相關內容