Managing HP X9000 Network Storage System Remote Replication Application Note (TA768-96049, November 2011)

replication and watches for errors and failures. The replication software uses file level checksums
at the source and target to confirm that replication was successful.
When a replication task is started, the source system sets up a replication stream from each file
serving node to the target file serving nodes. Each node replicates data from the segments it owns;
a node cannot replicate data from segments it does not own.
A source file serving node chooses a target node using a round-robin approach based on the
number of transactions it has completed on the current target node. A transaction is the transfer of
a file or a metadata change. There is no preference of target nodes in relation to the segments
they own.
A remote replication task includes the initial synchronization of the source and target file systems.
The synchronization takes place before replication is started and involves scanning the target
export and the replication source to determine the changes needed to bring the target into sync
with the source.
When using replication, be aware that:
• If a file or directory on the target export does not exist on the source, it will be deleted.
• If a file on the target has the same attributes and checksum as a file on the source, the file
does not need to be replicated. This leads to the possibility of seeding a target with files from
a source to avoid the network traffic generated by the initial synchronization.
If a replication task fails or is stopped and then restarted, the scan is repeated.
Configuration examples
This document includes three examples that show how to manage remote replication tasks. In the
examples:
• The source cluster is ibrix01 and the target cluster is ibrix02. Both clusters are X9320
systems.
• The DNS names for the Fusion Manager VIFs are ibrix01-fmgr and ibrix02-fmgr,
respectively.
• Nodes in cluster ibrix01 have hostnames ibrix01a and ibrix01b.
• Nodes in cluster ibrix02 have hostnames ibrix02a and ibrix02b.
• The source file systems on cluster ibrix01 are ibfs1 and ibfs2.
• The target file systems on cluster ibrix02 are also named ibfs1 and ibfs2.
• Replication takes place over bond1.
The examples are:
• Continuous replication of /ifs1 on cluster ibrix01 to /ibfs1/target1 on cluster
ibrix02. The examples use both the GUI and the CLI for this replication.
• Run-once replication of directory /ifs2/src1 on cluster ibrix01 to /ibfs2/target2
on cluster ibrix02. The examples use both the GUI and the CLI for this replication.
• Run-once replication of a file system snapshot of /ifs1/src1 on cluster ibrix01 to /ifs2/
target3 on cluster ibrix02. HP recommends using the GUI for this replication.
Configuring a remote replication export for inter-cluster replication
Before configuring an inter-cluster replication, as in the examples, you must create a remote
replication export. The steps are:
1. Register the source and target clusters with each other to form a replication pair.
2. Export the file system and, optionally, the directory on the target cluster that will receive the
replication. A target can be at the root level of the file system or can be a subdirectory. Target
6 Using remote replication