Hi We want to use Kafka Connect to copy data to HDFS (using kafka-connect-hdfs) in parquet format and was wondering if its a good idea to collocate distributed Kafka Connect 1-1 on Kafka Brokers?
Considering the parquet indexing process would steal (a lot of / too much?) computing resources from the brokers. The alternative is to deploy on separate hardware (1-1?). Experiences in this area? Cheers, -Kristoffer
