Kafka Consumer architecture design: java plugin or external client

I'm coding a solution for data consumer to a kafka cluster (using dockers for kafka). The point is that I know how to code a solution of a java plugin executed with kafka-run-class.sh or on the other hand developing an external client (e.g. a nodejs client consumer) I have first to take a decision about the way to take, in terms of extensibility and performance which would be the better solution.

This is for a ubuntu server, on a full dockerized environment, I use nodejs as a core language for web services and connecting modules...

Node js consumer (I'm still not tunning the kafka client):

var consumer = new kafka.Consumer( client, [{ topic: 'topicVehicle' }], { autoCommit: true });

I have to deal with the problem of maintain code in various layers (java, nodejs... probably python) and I'm wondering if it is better one solution of another (to avoid bottle necks and so on).