In our worker-distributed props, we have:
value.converter=io.confluent.connect.avro.AvroConverter
value.converter.schema.registry.url=<dev-schema-url>
We've noticed that the performance is bad when we use schema registry. Can I use AvroConverter without the schema registry? I tried https://github.com/farmdawgnation/registryless-avro-converter but we run into:
java.io.IOException Not a data file
Is schema registry contacted with every message the kafka connector reads?
Is the schema cached?
What are the other options to use AvroConverter without the schema registry?