Effective in version 6.1, the performance and scalability of applying retention policies have been improved in the following ways:
You can resume the application of a retention policy from the point of failure.
Previously, when you resumed the application of a retention policy, the job restarted.
The query construction for a retention policy is optimized. Child tables have a metadata column, either a tag or legal-hold column, that gets populated with the attribute applied to the parent. This allows the query to construct a join of just the child and immediate parent.
Previously, the query constructed a join that included the child, immediate parent, grandparent, and beyond. This strategy did not successfully scale to more than seven to eight levels.
Applying a retention policy at the application level and unrelated-entity level does not depend on the relationship between tables. Because of this, any relation-based search or navigation on unrelated entities is not supported.