I've been doing a simulation of the possible data ingestion flow where incoming messages are funnelled through Kafka and stored on ELK for fast analytical and intelligent insights into data quality. Instead of working with raw messages which require several stages of data transformations (format conversion, filtering, mapping to business attributes), I've taken post-processed files as initial input. However, the files are in AVRO format which is not an issue for Kafka but would be a real parsing disappointment for Logstash.