Today I was asked my opinion about whether we should buy another data migration product for heterogenous databases in addition to the two (2) we already have.
We have two (2) data migration products:
- One for pumping data from Oracle RDB to Oracle RDBMS
- Another for pumping data from Oracle RDBMS to Microsoft SQL/Server and to other Oracle RDBMS databases
Both of these products have a significant investment in human capital. In other words, people have taken years to get these products working to meet our needs. There is a lot of painful learning gone to make these products work.
Maybe we should consolidate on a single product, but the costs of doing so would be, I think, far greater than any savings in licence costs.
So, my opinion is that we should stick with what we have. From what little I have seen, the data migration offerings are not anywhere near the status of plug and play.
The biggest handicap is that a considerable amount of knowledge about what is in the database is needed to ensure that the data migrations work correctly. In other word, the semantics of the data is not reliably and consistently available inside the database.
Data migration is not simply a way of moving bytes from one platform to another. It has to preserve the information represented by that data as well. In other words, the problem is how to move data between differing data models? One way seems to be that of Semantic Computing.
From what I have read, semantic computing is still in its early stages of development. And when it matures, then there is the problem of how encode old data models to be usable in semantic computing.