Kafka is a distributed system and can be configured to provide a high level of resilience on its own. That is the reason people are in doubt and don't know if they even need a backup stream. We go over the different arguments and questions such as "What if somebody makes a human error which causes events to be deleted after a day?".
In this talk we explain how you can easily store your events in cloud file storage. We also explain how you can restore the data in topics of your own choice and how this relates to scheduled database backups in your environment.

Kafka operational best practice