August 2017
Beginner to intermediate
278 pages
6h 40m
English
Spark uses Kafka high level consumer API to implement receiver. This is an old approach and data received from Kafka topic partitions are stored in Spark executors and processes by streaming jobs. However, Spark receiver replicates the message across all the executors, so that if one executor fails, another executor should be able to provide replicated data for processing. In this way, Spark receiver provides fault tolerance for data.
The following diagram will give you a good idea about the receiver-based integration:

Spark receivers only acknowledge to broker when message is successfully ...
Read now
Unlock full access