Zero-copy Replication for ReplicatedMergeTree over HDFS Storage

Why it matters

This feature addresses performance and storage overhead issues during replication by eliminating redundant data copying processes. It allows replication to be faster and more storage-efficient, especially beneficial for large datasets stored in HDFS.

How to use it

To use zero-copy replication, configure your ReplicatedMergeTree tables to store data on HDFS and enable the zero-copy replication option. The replication mechanism will then reuse existing data files on HDFS during replication processes automatically.