What every software engineer should know about Apache Kafka

Re:What every software engineer should know about Apache Kafka by epistasis on Saturday May 16, 2020 @05:46PM (#23207477)

The point of that is for people trying to figure out why streams are a useful abstraction at all what's needed to make them useful are some sort of aggregation, and of course tabular state is a common end point.

The article does not recommend writing this code yourself, it shows how to aggregate data into usable forms.

So I think your concerns may be a bit overblown. If you think that ksqlDB or Kafka Streams, the tools shown in this blog post, are are at risk for what you warn, this comment would be a valid criticism. But it's clear that the article isn't advocating for people to write their own versions of that...

[ Parent | on HN ]

Re:What every software engineer should know about Apache Kafka by SpicyLemonZest on Saturday May 16, 2020 @05:59PM (#23207568)

Yes, they're definitely at risk. ksqlDB does not appear to have transactions at all.

[ Parent | on HN ]

Re:What every software engineer should know about Apache Kafka by camdenreslink on Saturday May 16, 2020 @09:23PM (#23208642)

The transaction boundary is the message/event itself (in the case of an event sourced system, which is what this would primarily be used with). Everything required to understand some event is contained within the payload. It is the atomic unit in this type of system.

[ Parent | on HN ]