This is bad advice, taken on its own.

Denormalization is a species of optimization.

Don't blithely do or recommend denormalization without specific data supporting the actual, documentable performance issue. If you don't know the actual issue, you might well simply make things worse.

This is bad advice, taken on its own.

Denormalization is a species of optimization.

who recommended denormalization?
i said, *if* you tend to not normalize because of performance issues, use a search engine instead.

Howdy!

"normalization can lead to worse performance" is a useless statement. Normalization can also lead to better performance. It *does* lead to more robust data models that more clearly communicate their design.

Deciding not to normalize amounts to denormalization.

Making that decision for "performance" implies that the normalized form has performance issues. Without normalizing and testing, you cannot properly assess the differences. You end up making assumptions about where to optimize, and those assumptions tend to be dead wrong too much of the time.

Further, what do you mean when you say "consider using a search engine"? The data being described seems to be live data that will be updated transactionally. That need points toward more fully normalized forms, but it is not clear how the balance between update activity and query activity lies. A "search engine" is no help for modifying the data, and may be counterproductive if you have to frequently query rapidly changing data.

yours,
Michael