Data migration

Data migration moves selected data from an existing system to a new system or data model. It can include cleansing, transformation and mapping new identifiers. A migration has an agreed scope; not every historical record automatically has to become part of the new working data set.

In a fictional example, the old customer K-72 becomes C-910 in the new system. Orders included in the migration must refer to that new customer identifier. Having the same number of customer rows after import does not establish that these relationships are correct.

The transfer needs a cutover point

If the source system keeps running, an initial transfer may be followed by applying changes made in the meantime. AWS DMS describes, for example, an initial full load followed by change replication. Your own process needs to define when the old system stops accepting changes and when the new system takes responsibility.

Check field meanings, relationships, rejected records and pending changes before cutover. Also define how a failed transition will be handled. An “import complete” message does not replace these checks or authorise deletion of the old system.

Sources

• AWS — Working with DMS tasks