Question

Almost in all stages of my application, the database inserts / updates / deletions are done by the application code (in my case, PHP).

However, in some situations, I am using MySQL triggers to do some calculations and changes to the tables.

Is this a recommended approach? Or is there any best practice rule that only the application should do all the work and not the database?

Thanks for any recommendation.

Answer 1

The database is typically the hardest thing to scale. As a result, at very high scale, one popular strategy is to use only the most basic database features to ensure maximal database performance — which means no triggers, no foreign keys, no joins (accomplished via denormalization), etc. However, this is only something you have to worry about if you’re in, say, the top 1% of the web.

However, simplicity is always a high priority as well. And triggers are often the simplest solution. If you’re not going to be anywhere near the top 1% of the web in the near future (and let’s be honest — few, if any of us here are) go ahead and use triggers.

Designing an application for facebook-levels of scale requires far more work and maintenance overhead than writing an application the simple way. And if you don’t reach facebook-level traffic, all that work was wasted.