On: Surviving Django, if you care about databases

[ensignavenger]

I support a very large database, containing product data for a major US retailer. If I understand the history of this database correctly (I am not directly on the team that manages it, but my team supports them in various ways) the database began life as an MS SQL DB. Many years ago it was migrated to MySQL. Now, they are in the process of migrating it to Postgres. So yes, migrating data to different DB's does happen, especially when you consider an applications life over decades.

[kevindong → ensignavenger]

Why was the underlying database changed?

The most obvious reason I can think of being licensing concerns with both MS SQL and MySQL.

[ensignavenger → kevindong]

The first migration from ms sql to mysql was licensing costs, and the desire to host on Linux. The more recent plan to move to postgres is motivated in part by features and in part by standardizing on postgres.

[grey-area → ensignavenger]

Features that you wouldn't be allowed to use if you try to remain database agnostic.

DB migrations happen, but not often enough to use a lowest-common-denominator db agnostic library and try to remain compatible with all systems.

[ensignavenger → grey-area]

Well, that is another area the article got partially wrong. There are many postgres specific features included in Django, and other available via third party packages. And for most things that the ORM doesn't specifically support, it gets out of the way and lests you do them anyway, by executing your own custom SQL, right from the ORM.