In DBZ-2575, a new internal property, internal.snapshot.scan.all.columns.force, was added. When enabled, it reduces the overhead of reading column details by performing a per-table column metadata fetch rather than creating a cursor for all tables within the schema.

The confusing part is that this property must be prefixed with database. so that when we use the Configuration#subset method to get all properties that are prefixed with either database. or driver., they're included in the JdbcConfiguration.

So users need to specify this property as:
database.internal.snapshot.scan.all.columns.force

Would it not make more sense to adjust how the JdbcConfiguration is created by having the subset call include all properties that are prefixed with:

• database.*
• driver.*
• internal.database.*
• internal.driver.*

This way the configuration property in the class that be defined much like how we define the USER, PASSWORD, and other relational JDBC properties, so that the usage is better self-documented rather than needing to understand how the value gets passed in through our various classes?