KIP-64 — Allow underlying distributed filesystem to take over replication depending on configuration
Discussion Broker
KIP-64 proposes a broker config to delegate replication entirely to the underlying distributed filesystem (e.g. NFS/SMB-mounted HDFS or GlusterFS), treating a successful `flush()` as a signal that all ISR replicas have the data and disabling Kafka's own replication traffic. This eliminates redundant data copies and inter-broker replication overhead when the storage layer already provides its own fault-tolerant replication.
Details
| Author | Arvind Kandhare |
| Status | Discussion |
| JIRA | KAFKA-3867 |
| Wiki | View on Apache Wiki |
| Created | 2016-06-18 |
| Last Modified | 2016-06-19 |
Explore how this KIP affects the Kafka protocol in the Protocol Explorer, or see the full KIP database.