According to the redshift docs, a column defined with VARCHAR(N) can store up to N bytes of text. In other db platforms (for example in MYSQL), VARCHAR(N) can store up to N characters. This subtle difference leads to incorrect table DDL when using Redshift as a destination and leads to strings that are too long for the destination table when the source database uses character sets that allow more than 1 byte per character.

For example, I set up a connection using MySQL as a data source, and Redshift as the destination, with a text column defined as `varchar(60)` using a `utfmb3` character set (allows up to 3 bytes per character). In this case, fivetran creates that column as `varchar(60)` in Redshift when it really should be creating it as `varchar(180)`. Similarly, a `utfmb4` character set in MySQL should be mapped to a `varchar(240)` column in Redshift.

I haven't tested this on other source DB types but I noticed this when writing custom ETL code and was curious if it was happening in Fivetran as well.