Understanding Reliability with Kafka
How partitions enable work not only to be distributed, but reliably distributed ! What happens if the disk fails or the data is inaccessible ?
Reliability with Apache Kafka Replication
There is no redundancy between the nodes, there could be unrecoverable data loss because the data being managed by the failed broker, unless replicated to another cluster, is now accessible.
The property “Replication-Factor” will enable to handle failed broker or server.
kafka/bin/kafka-topics.sh --create \
--zookeeper localhost:2181 \
--replication-factor 3 \
--partitions 3 \
--topic unique-topic-name
The replication set to 2 or 3 at minimum so that failures or machine maintenance will not interrupt the cluster’s operation.
Replication with a factor to 3
If one broker is down, Kafka search for a new following peer to replace the quorum member.