It is important to have a process for backing up your environment so that
you can easily recover from a storage device failure or loss of data. In SAN
File System, you must save both the file data and metadata together when you
back up the global namespace. These are used to recreate your user data. You
must also backup the system metadata, which is used to recreate the SAN File
System configuration.
SAN File System does not provide backup and restore functionality; instead,
it supports backup tools that are already present in your SAN environment.
Depending on the type of failure, you might need to restore a single file,
an older version of a file, a directory, a volume, or the entire system. SAN
File System supports various options for protecting the system, including:
- Creating FlashCopy® images
- Backing up files using third-party backup and restore applications that
are already present in your environment, for example IBM® Tivoli® Storage Manager (TSM), Legato
NetWorker, or VERITAS NetBackup
- Using copy services that exist in the underlying storage device (for example, FlashCopy and
Peer-to-Peer Remote Copy (PPRC) functions in the IBM TotalStorage® Enterprise
Storage Server® or IBM TotalStorage SAN Volume Controller
- Saving the cluster configuration (system metadata)
There are two basic methods available for backing up and restoring your
data:
- File method
- Volume method
The file-based method saves and restores data at the file level. It uses
the FlashCopy function
or other third-party backup and restore application in your environment to
back up or restore your user data. Use the file-based approach when files
have been lost but the overall system remains healthy.
The volume-based method saves and restores data at the device level (that
is, a "just-a-bunch-of-bytes" approach). To adopt the volume method,
however, the actual copying and restoring of data must be provided as a service
by the underlying storage subsystem. Use the volume-based approach when disaster
strikes and the system, as well as the FlashCopy images, are unusable.
Tip: Your backup and restore process does not have to be centralized
and homogenous, covering the entire SAN, although such a process simplifies
the procedure. You can use the volume method even for a fragmented SAN that
requires a piecemeal volume copy across two or more storage subsystems. In
such a scenario, you would be responsible for manually managing those multiple
backup sets as though they were a single backup set.