In my understanding, every liquibase update run
- goes through the complete changelog file
- for every changeset met it calculates it's MD5 sum value
- and for every changeset checks it against DATABASECHANGELOG table to see whether changeset already deployed or modified
If imagine the large number of changesets already deployed, for example, tens of thousands or more, the process of adding just a single changeset might seem (at least theoretically) time consuming.
What is the correlation between the number of changesets and "migration" time? Are there any limitations? If yes, what are the possible alternative solutions for database migrations with big amount of objects and changes?
