I have been working on data migration recently. I thought data migration would be a simple task, but I immediately dispelled this idea after doing so, this is a very complex task and the work content is not easy.
Several issues to be considered before migration
- The correspondence between the migration source and target.
- Division is a wise choice.
- How to verify the migrated data.
- Which types of tools are required to support the migration process.
- Initialize the verification scenario data after migration.
Migration tools
The migration tool uses kettle to complete scenarios that require complex logic conversion, in scenarios where the structure logic is simple and does not involve data analysis and transformation, but the structure is large, we choose to build our own toolkit.
Migration
Summarize the problems we encountered during migration,
- The most time-consuming verification process after migration is not communication/communication, because we migrate the business system, after the migration, a lot of sub-systems need to be re-initialized. If these jobs are independent, they can also be made into a large toolkit.
- There is no way to roll back. After the migration is completed, the business process is implemented across multiple subsystems, which makes everyone passive when we do not propose the idea of dividing the task into stages, seriously practice the idea of breaking down big problems.
- In verification scenarios, the Business Process of the migration source and target application scenarios is consistent, but the implementation is technically different, it involves a wide range of business knowledge, so we need to coordinate multiple businesses familiar with different subsystems and developers for joint verification, but this will be very costly.
- The demand changes in the migration process, which once again proves that as you learn more about a tool, there will always be new ideas and suggestions for improvement.