spark cassandra connector write timeout when when tuning write conf

I have noticed a very low throughput in our pipeline when inserting data from spark to cassandra (less than 1 MB/s per core). When trying to tune write conf (spark.cassandra.output.concurrent.writes, spark.cassandra.output.batch.grouping.key and spark.cassandra.output.batch.size.rows) I am getting quickly a write timeout.

My questions:

• Is it recommended/normal to increase the cassandra write timeout when writing data in batch (via spark)?
• Is it possible to increase it just for spark workloads? or just for batch writes?
• The default value for spark.cassandra.output.batch.size.bytes is 1024, I find that too low as a default value, I guess in most times that would correspond to 1 or 2 rows, am I missing something ?

I am using spark-cassandra-connector 2.4.3