Using the Kafka Extensible Common Data Framework for Fault Tolerance
To ensure fault tolerance, you can set up multiple instances of s-Server to listen to each partition, using a program like Puppet to start servers and pipelines. You can run more than one pipeline on each version of s-Server, but you cannot have multiple pipelines on the same server listening to the same partition.
The diagram below shows multiple instances of s-Server listening to Kafka partitions. This prevents data loss if an s-Server instance goes down.