The replication service uses the following strategies together to reduce the amount of data required for asynchronous replication:
Incremental asynchronous replication
Source and target clusters are usually deployed at two geographically distant sites, and the bandwidth between them is usually limited. To fully utilize bandwidth and reduce transmission time, if the source cluster and target cluster have completed an asynchronous replication job before, subsequent jobs use incremental asynchronous replication to reduce the amount of data transmitted.
During data transmission, the target-side replication service divides replicated data into blocks and calculates fingerprints for the blocks. The fingerprints are stored as metadata in the target-side replication service. During incremental asynchronous replication, the target-side replication service queries all fingerprint data in the previous replication chain, and compares the fingerprint of each data block from the previous replication with the currently calculated fingerprint of the data block. If the fingerprints of a data block are identical, the data block has not actually been updated and is skipped during replication. This mechanism ensures that only incremental data that has actually changed is transmitted and stored during replication jobs, improving replication efficiency.
Skipping unallocated data blocks
When a volume uses a thin provisioning storage policy, the replication service queries ABS volume metadata to determine whether unallocated data blocks exist, and skips all data blocks that are not actually allocated during replication to reduce the size of replication files.
