Kafka data loss scenarios
Kafka topic partitions are replicated across brokers. Data loss happens when the brokers where replicas are located are unavailable or have fully failed. The worst scenario — and where is no much to do — is when all the brokers fail; then no remediation is possible. Replication allows to increase redundancy so this scenarios is less likely to happen.
The following scenarios show different trade-offs that could increase the risk of lossing data:
Read more