Kafka, Spark, and Elasticsearch (architecture comparison)

	Kafka	Spark	Elasticsearch
Horizontal Scale	Topic/Partition	RDD/Partition	Index/Shard (default 5 primary shards per index)
Resource/cluster manager	Zookeeper	YARN, Mesos, or standalone	Master Node
Cluster node	Broker	Worker	Node
Execution unit	Partition Leader	Executor	Lucene Index
Core		RDD	Inverted index
Applications	Producer/Consumer	Driver	Search
Replication	Follower (ISR)		Replica shard

Kafka:

http://cloudurable.com/blog/kafka-architecture/index.html

Spark:

https://www.dezyre.com/article/apache-spark-architecture-explained-in-detail/338

https://www.edureka.co/blog/spark-architecture/

Elasticsearch:

https://buildingvts.com/elasticsearch-architectural-overview-a35d3910e515

http://solutionhacker.com/elasticsearch-architecture-overview/