Policy-Based Data Replication
System Administration Manual
332
Policy-Based Data Replication
Titan SiliconServers supports a policy-based data replication. This allows administrators to
setup and configure replication jobs independently from other backup strategies using the Web
Manager.
Titan SiliconServers can be configured to perform incremental data replications. When a
replication policy is first set up, the SMU performs an initial copy of the source volume (or
directory) to a target destination. After the initial copies are successful, incremental copies are
performed at the scheduled intervals. Incremental block level replication can optionally be used
to replicate large files more efficiently. During an Incremental data replication, files that have
been changed since the last scheduled replication will be replicated to the target in full. With
incremental block-level replication (additional license required), only the changes in files are
replicated and not the whole file, thus reducing the amount of data replicated. This reduces the
overall replication time.
During replication configuration, a replication policy is setup to identify a source File System,
Virtual Volume, or directory, the replication target, and a replication schedule. Pre-replication
and post-replication scripts can also be set up during that time. Replication rules are then
defined to include various optional settings. These optional settings are explained later in this
section.
Incremental Data Replication
The Titan SiliconServer provides support for Incremental Data Replication (IDR). IDR is
performed under control of the
System Management Unit
(SMU). IDR uses the same data
management engine as
NDMP
to copy the contents of an entire file system, a
Virtual Volume
, or
an individual directory tree to a replication target. The section
Performing NDMP Backups
includes information about NDMP, much of which applies to IDR. Replication is performed
incrementally, which means that after the initial copy, only changes in the source volume or
directory are actually replicated on the target. Snapshots are used to ensure accuracy of the
replication.
Note:
In case the snapshot taken prior to the previous replication is lost, the
full data set is replicated.
Once a replication policy and schedule is set up, the IDR process will take place automatically at
the specified interval. The replicated data can be left in place (and used as a standby data
repository). In addition, the replicated file system or directory can be backed up through NDMP
to a Tape Library System (TLS) for long-term storage (which can be automated).
The replication target should not be actively used during the replication process. Accessing files
during the replication process could prevent those files from being updated. Also any changes
applied directly to files at the target might be lost when the file is next updated from the source.
IDR supports the following targets for replication:
