A Deep Dive into Apache Kafka This is Event Streaming by Andrew Dunnings & Katherine Stanley

This session dives deep into the architecture of Apache Kafka and how to use it in practice. Kafka offers a new take on publish/subscribe messaging that is very different from other messaging systems. But Kafka isn't just a messaging system - it's an event streaming platform.
Become an expert on Kafka concepts such as topic partitioning, consumer groups and exactly-once semantics. Find out best practices for using the APIs to achieve performance and reliability.
Learn about advanced concepts such as stream processing using the Kafka Streams and integrating with other services using Kafka Connect.
Katherine Stanley is a Software Engineer in the IBM Event Streams team based in the UK. Through her work on IBM Event Streams she has gained experience running Apache Kafka on Kubernetes and running enterprise Kafka applications. In her previous role she specialised in cloud native Java applications and microservices architectures. Katherine has co-authored an IBM Redbook on Java microservices and has contributed to the open source microservice project Game On. She enjoys sharing her experiences and has presented at conferences around the world, including the Kafka Summits in New York and London, JavaLand in Germany and JFokus in Sweden.
Andrew Dunnings is a Software Developer in the IBM Event Streams team based in the UK. Andrew joined IBM as a graduate in 2017 and has developed expertise in running Apache Kafka on Kubernetes through his work on IBM Event Streams. He has experience introducing IBM customers to Apache Kafka and is keen to share his knowledge of both Kafka and Kubernetes with the developer community.

[@SportAnalysisKing]

kafka connect starts at 59:42

[@AndrewBourgeois]

14:30 "min.insync.replicas" won't prevent a Producer from putting messages into Kafka unless "acks=all" is used in the Producer configuration.

[@wallstreetdonut6539]

I agree with Andrew and just for someone who wants to know why, "min.insync.replicas" should not be the size of the total replicas if "acks=all". It's because if one of replicas doesn't work, then the number of replicas which could be working will eventually be less than the total number of replicas so that "min.insync.replicas" couldn't satisfy "acks=all". Programs will stop.

[@davidlester6673]

Big yes to removing zookeeper. Thumbs up.

[@SpiritOfIndiaaa]

very nice , thank you , but need to more info on prod tuning of kafka.

[@bawaparampreet24]

no two consumers in one consumer group can get same message. presenter was wrong at 43:41. if there are more consumers than partitions others would sit idle

[@Ghaleon]

Anyone recommend a good place to learn more about kafka and/or where community can talk about kafka? I'm using node with kafkaJs so far.
This video is really helpful. thank you guys

[@cliffmathew]

Quite useful.

[@NishitVishwanath]

Could you please share the slides ?

[@miguelhenriquez2050]

at 27:32 he says "back to producers", but he meant "back to topics"

[@danyald4880]

Very nice

[@foodaupair]

1:11:00 docker

[@enzocisternino3709]

INSANE !

[@AhmetMurati]

Nowadays Kafka is built in Scala

[@SportAnalysisKing]

8:00 starting topics

[@amotorcyclerider3230]

Hot!

[@vishalchopda6169]

Hisgsiwgoai future

[@djcb4190]

ENglish.

[@digisecureagent7679]

This video is 3 years old and is not relevant anymore. Use KSQL instead

[@lhxperimental]

Thank you. Such pointers really help.