On the other hand, tape backup is usually a host-based process that copies data at the file level using software installed on the host; the backup server takes care of tracking what has been backed up via catalog or database. Restoring a full backup from tape to a remote array must be done using a host that has a file system and directory structure on that array to restore at the file level. While this will get the data on the remote array, it does not produce a "block map" that the storage array can understand and use for future replication.
There are limited options to get this done and they include:
- Install the target (remote) SAN in the same location as the source SAN to get the initial baseline replica and then move the target SAN to the remote location to resume the replication process for changed blocks only.
- Work out an agreement with your carrier to temporarily have access to additional network bandwidth between both sites to complete the initial replica. Note that this may still take a long time depending on the bandwidth available. For example, 21 TB of data could take around 20 days over a 100 Mbps link. You can also take advantage of WAN optimization technology to help reduce the amount of redundant data being sent across the WAN link.
This was first published in November 2010