The advantages of normalization:
Normalization of update operations is usually faster than inverse normalization.
When the data is well normalized, there is little or no duplication, so you only need to modify less data.
A normalized table is usually smaller and better placed in memory, so it is faster to perform the operation.
There is little extra data that means fewer distinct or group by statements are needed to retrieve list data.
Disadvantages of Normalization:
Canonical design schemas often need to be associated. A slightly more complex query statement may require at least one association, perhaps more, in a schema-conforming paradigm. This can be expensive, and may invalidate some index policies. For example, normalization might place columns in different tables, and those columns can have data in the same index in the same table.
Advantages of Inverse paradigm
Can be very good to avoid association.
If you do not need the associated table, the worst case for most queries, that is, the index is not used, full table scan. This can be much faster than association when the data currency is large, thus avoiding random I/O.
In the real world, it is very rare to use the paradigm or inverse paradigm of schema. Instead, it is possible to use partially normalized schemas, cache tables, and other techniques. The most common way to reverse-normalized data is to copy or cache the same specific columns in different tables.
This article from "Sharing learning" blog, declined reprint!
The advantages and disadvantages of paradigm and inverse paradigm in MySQL