One of the hard nuts in a typical software project is the data migration.
A typical situation for the migration team:
- The data should be migrated (integrated etc.) from the old system to the new one
- The data model of the old system is not known precisely any more (optionally: the old provider does not even exist any more, or, its involvement requires a heavy investment)
- In course of the time certain fields on the screens were not used according to standards (e.g. storing a GLN number in an address field)
- The data model, used for mass data input/output of the new system is partially known (sadly a reality with some software companies still today)
One of the possibilities is the usage of software robotics (RPA – Robotics Process Automation). How?
- The screens/reports of the old application are still used and known very well by the business specialists
- The screens of the new system are also well known and usually designed together with the same business users who use the old system
- With software robotics it is possible (and usually not very complex) to build an algorithm that opens the old screen and does copy-paste into the new system
- This can be combined with some data transformation as well (e.g. cleansing of addresses etc.)
In our experience the above migration method is surprisingly fast both in terms of business analysis and can also be used to migrate multiple millions of records (degree of parallelism).