Which one to use - Apache Kafka versus RabbitMQ

Best case scenario (use case)  for Apache Kafka

1) Requirement for partitioned order at least once write. Requirement for  A to B streaming with high throughput.  

2) Durable message store . Why would this be important? Consumers can get a replay of the event stream - whereas limited in message brokers - where once message is delivered , removed from queue

3) If a solution requires data stream processing.  

4) Event sourcing required. Event sourcing persists the state of a business entity e.g maintaining a chain of event at different states in a sequential order for a Customer

5) More components in Kafka - (once you get your head around it)  , supporting great flexibility. There is more work to get Kafka operating - in comparison to RabbitMQ  - but once operating has plenty of options, for developers and operations

6) High Availability . Both  RabbitMQ and Kafka have High Availability - but Zookeeper offers the cluster management component and is also Highly available 

7) Kafka exploits sequential disk IO , therefore high throughput and less hardware. I'd class this as a supporting factor. 

Best case scenario (use case)  for RabbitMQ

1) Easy to setup in comparison to Kafka

2) Work with a combination of existing protocols - AMQP 0-9-1, STOMP, MQTT, AMQP 1.0.

3) Finer grained consistency controls

4) Wild card event subscription - RabbitMQ supports wild cards for event subscription

5) Pivotal Cloud Foundry (PCF) compatability . To highlight compatability - it takes a one-line command to set up &  provision RabbitMQ

So this blog post is about Apache Kafka and RabbitMQ. As a DBA - I've found knowing more about messaging\queuing\data streaming  technologies assists working with  development teams

Let me know other  factors for working with Kafka and RabbitMQ