Data migration

A data migration is a one-time move of data between storage systems, usually driven by a hardware refresh, a consolidation, or a move to the cloud. It has a start and an end, which is what separates it from mobility.

Key takeaways

A data migration is a one-time move between storage systems with a start and an end.
Planning starts with an index of the source, so only what is needed moves.
Chain of custody and a warm cutover reduce risk and downtime.

Why do unstructured data migrations fail?

Usually for reasons that start before the move:

Nobody knows what is being moved, so waste is copied along with valuable data.
Permissions and owners break on the new system.
Applications point to old paths that no longer exist.
The schedule is wrong, because size and file counts were estimated, not measured.

An inventory taken before the move prevents most of these problems.

How do you plan an unstructured data migration?

Start with an index of what is on the source system, then decide what moves at all. A typical plan has four steps:

Index the source.
Leave behind duplicates and data with no remaining use.
Preview the move as a dry run.
Move the rest in batches, with each file recorded in the catalog.

Planning from the index usually makes the migration smaller, because part of the source never needed to move.

What is a dry run, and why does it matter before a move?

A dry run reports exactly which files a rule would affect, without moving anything. It turns a policy from a risk into a reviewable proposal, and it lets someone approve a change covering millions of files with a clear conscience.

What is chain of custody in a data migration?

Chain of custody is the record that shows every file arrived at its destination complete and unchanged, and who moved it when. In practice it combines a list of files before the move, checksums compared after the copy, and a log of each step. It matters most for regulated, legal, or research data, where an organization may have to prove that a file is the same one that was created.

What is a warm cutover in a data migration?

A warm cutover is a migration in which the source stays in use while data is copied, and a final short sync moves only what changed before users switch to the new system. It reduces downtime compared with a cold cutover, where the source is frozen for the whole copy. The trade-off is planning: changes during the copy must be tracked so the final sync is complete.

What is data repatriation?

Data repatriation is moving data and workloads from public cloud back to on-premises or colocation storage. It usually follows a cost review where read and egress charges turned out to be higher than modeled. The same catalog that justified the move out is what makes the move back measurable.