Kafka + Kafka Streams

182 views
Skip to first unread message

Kishan Patel

unread,
Sep 19, 2016, 6:39:50 PM9/19/16
to DDD/CQRS
I am working on implementing a event sourced, cqrs, microservice based system. 

We are planning on using Kafka as our Event Store (the topics being our aggregate event log) and Kafka Streams as our view updater service. This post on confluent.io explains the approach. We would be writing to an external database instead of making use of the local state provided by Kafka streams. We have no experience with building these sort of systems - so would be great to know - any possible flaws you can see with the approach, things we should look out for, drawbacks of using Kafka as a Event Store. Any insights would be very valuable as we haven't gotten too far with the implementation. 

Thank you!

Chris Richardson

unread,
Sep 19, 2016, 6:51:23 PM9/19/16
to ddd...@googlegroups.com
While I really like Kafka, I am a little puzzled by the confluent.io article.
Specifically, if you want to "load" a particular aggregate from the Event Store then you need to retrieve the events for that Aggregate instance  by primary key.
Kafka doesn't efficiently scale with a topic per aggregate instance so its not clear how you can implement the "load" operation.

Chris
-- 
Learn microservices - http://learnmicroservices.io
Microservices application platform http://eventuate.io


--
You received this message because you are subscribed to the Google Groups "DDD/CQRS" group.
To unsubscribe from this group and stop receiving emails from it, send an email to dddcqrs+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.



Reply all
Reply to author
Forward
0 new messages