DBZ uses Kafka Connect's (KC) Schema.java and Struct.java as its data model when converting data from databases. This creates a hard dependency on KC, which has a few side effects:

1. It's a bit irritating to people using DBZ without Kafka.
2. KC's schema and struct are limited in some ways (e.g. no union, no aliases, no self-referencing, no constraints, etc).
3. Lack of control over DBZ's data model-subject to changes in Kafka.

There are probably a few other drawbacks as well; this is just the top of my head.

As per- the discussion in DBZ-6234, I think we should start discussing how to separate DBZ's model from Kafka Connect.