Policy-based data management

Policy-based data management is managing data through written rules rather than one-off projects. Each rule names a set of files by attributes, such as project, age, owner, or retention class, and states what happens to them, such as a move to an archive tier. The rule runs on a schedule, previews its effect as a dry run, and records what it did. The result is a lifecycle that keeps running after the first cleanup ends.

Key takeaways

Written rules replace one-time cleanup projects with ongoing decisions.
The catalog selects files by attributes, and a data mover carries out the move.
The same rule can work across every storage vendor.

Why use policies instead of one-time cleanup projects?

Because cleanup projects end and data keeps growing. A one-time project reclaims space once, then the same problem returns within a year. A policy runs on every scan, applies the same rule every time, and records what it did, so the result lasts and can be shown to an auditor.

What is an example of a data management policy?

A media company writes one rule: every Sunday night, files in project folders marked as delivered, not opened in 180 days, and larger than 1 GB move to archive storage, and the project owner receives a list of what moved. The rule runs every week without a ticket, and every move is recorded, so anyone can see why a file left fast storage.

Is tiering a type of policy-based data management?

Yes, and it is the most common one. A tiering policy decides which files move to which storage tier and when, based on attributes such as age, last access, or project status. Policy-based data management covers more than tiering, though: the same kind of rule can apply retention, delete data that has expired, tag files, or copy a selected set to an AI pipeline.

How does a catalog decide what to move?

It applies rules to attributes it already holds. A typical rule combines age, last access date, path or project, owner, and format, then produces a list of matching files. The rule is written once and keeps running, which is what distinguishes it from an annual clean-up.

What is catalog-driven orchestration?

Catalog-driven orchestration means the metadata catalog decides what moves, and a data mover carries it out. A policy selects files by attributes in the catalog, such as project, age, or owner, and sends that list to a mover, which copies or moves the files to the right tier, site, or cloud. Because selection uses the catalog, the same rule works across every storage vendor.

What is the difference between automation and orchestration?

Automation makes one task run by itself, such as a rule that moves cold files to archive storage every week. Orchestration coordinates several automated steps, often across different systems, so they happen in the right order: find the files, tag them, copy them, check the copy, notify the owner, and only then remove the original. Automation saves effort on a task, and orchestration makes a whole process reliable.