Using rabbitmq instead of Kafka as a message broker

[warri...@gmail.com]

Hi, is it possible?

[warri...@gmail.com]

Or, alternatively, after reading binary log records from Mysql, redirect it Rabbitmq instead of Kafka

On Wednesday, January 25, 2017 at 4:16:09 PM UTC+2, warri...@gmail.com wrote:

Hi, is it possible?

[Emmanuel Bernard]

RabbitMQ AFAIK does not offer some key features of Kafka that Debezium expects:

- being able to reread a topic from the beginning of time (or something equivalent)

- offer total ordering (per partition)

- at least once guarantee (RabbitMQ might in some config, I don’t know)

- let the consumer control the index at which it has read vs be the one controlling via the broker

- let the consumer be super slow or even be offline without affecting the rest of the infrastructure

So that would not be a great fit.

--
You received this message because you are subscribed to the Google Groups "debezium" group.
To unsubscribe from this group and stop receiving emails from it, send an email to debezium+u...@googlegroups.com.
To post to this group, send email to debe...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/debezium/2c428354-8586-4a7b-9cdd-a44cc384e81d%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Randall Hauch]

As I mentioned in our dev channel on Gitter, the best that could probably be done is to embed a connector inside a RabbitMQ producer, which would have the responsibility of consuming the data change events, pushing them into RabbitMQ, and marking them as consumed when they have indeed made it into RabbitMQ. See http://debezium.io/docs/embedded/ for some background about using the older embedded engine, but https://issues.jboss.org/projects/DBZ/issues/DBZ-180 for a dramatically improved embedded engine. 

To view this discussion on the web visit https://groups.google.com/d/msgid/debezium/1ADC831D-E18B-4EFC-ADD5-BE1C33784D35%40hibernate.org.