NDMP Dump Backup Using Tree Walk

This article explains how to use the tree walk mode.

Dump time between Phase I to Phase II can take hours to complete due to large inode and directory counts.
If a backup environment has a large inode and directory count, the time between Phase I and Phase II can be extremely long. This is due to how DUMP will retrieve the files that will be backed.
By default we perform an 'inode walk', meaning it would reference the volume level inode table and entries. This will make DUMP go through the entire table, which can and will take longer, the more inodes that exist within the volume.
The 'tree walk' or 'logical path find' can be faster as it allows DUMP to reference only the root volume of the qtree and /or subdirectory for files to backup.
If the customer's backups and storage controllers are already configured to qtree backups, changing to 'tree walk' can be ideal for better performance.
However, if the customer environment does not have qtree configurations within his volumes, changing to 'tree walk' will not be efficient. The customer would have to re-construct the volume to contain qtrees and then move their data accordingly. This would potentially require a change to the policies in their backup application. Although they could benefit from the change, this would require a substantial amount of changes to be made within the volume and/or volumes first.