Object-based replication operations are based on snapshots.
The first time a replication is performed, a snapshot is taken (the initial snapshot), and the first replication operation replicates all objects on the source to the target. All following (incremental) replications take a snapshot of the changes to the file system and replicate only the objects that have changed.
Detecting and copying objects from a source to a target requires fewer system resources than detecting files and directories (which include directory structures and metadata). Object-level replications detect and replicate only those objects that have changed on the source file system, thereby using minimal system resources. Object replication is the fastest method for performing replications using the NAS server storage system.
In an object replication, a snapshot of a file system is replicated to another server, typically remote, to provide backup and recovery of the source data. The replicated files are immediately available for use in a disaster recovery situation. Additionally, the roles of the source and target servers can be reversed, allowing the target server to quickly take over the responsibilities of the source server.
Object-level replication has the following benefits:
- Higher performance than file-based replication. The greatest performance improvements are seen with incremental replication, especially dense file systems (many small files) or those file systems with a high rate of change. Larger file systems will achieve even greater improvements in incremental replication performance than smaller file systems.
- Object replication enables the ability to quickly failover in the event of a disaster.
- Object replication maintains the replication status on both the source and target file systems. If the replication relationship is broken, such as during a system shut-down or move, when the relationship is re-established, incremental replication can continue, rather than requiring a full re-sync of the file system.
Object-level replication has the following limitations:
- Object replication works at the file system level only; entire file systems may be replicated using object replication, but individual files or directories cannot.
- The target for object replication is read-only by file-serving protocols.
- During disaster recovery failover, target file systems are not accessible until promoted to primary. As the file system is being replicated as its constituent objects, the file system is in an inconsistent state until all objects have been replicated.
- CNS tree structures are not replicated; they must be manually replicated on the target system if CNS is used with object replication.