What role is Kafka playing in this infrastructure? Briefly motivate your answer.

Suppose that the latest data ingested to the HADOOP cluster were completely destroyed. How would you recover those data?

Describe how you might perform offline training within this infrastructure.

What technology or tool is required to retrieve data from the databases shown in the image.

What role is the Pub Sub component playing in the diagram, particularly as it relates to scalability.

Why is Flink required additionally to Kafka in this architecture?

Suppose you want to write data to Hadoop in Parquet format and the cluster is implementing _{at least} once semantics, what property is necessary in the Parquet connector?

 

Rider App Driver App API / Services Dispath Mapping & Logistic PRODUCERS Schemaless MySQL Cassandra DATABASES Kafka Realtime Pipeline Batch Pipeline Pub Sub Flink ELK Hadoop Mobile App Alerts, Dashboards Real-time Analytics, Debugging Applications Data Science Ad-hoc Exploration Activate Analytics Go to SettinReporting Windows.