As an example, sys.schema_unused_indexes is only available in MariaDb starting from 10.6…

I recommend making an index invisible for a fair amount of time before removing it. As we explain in the stream there is always a chance that the index is used on something that runs rarely such as a quarterly report. And note down the output from SHOW INDEX FROM <index-name> someplace just incase.

It’s not unusual in my experience that some tables have a lot of indexes, sometimes these are bulky and can add up to even greater than 100% of the size of the table.

Duplicate indexes and unused indexes is a bit complicated if you consider unique keys. A unique key may be needed even if it counts as a duplicate or an unused index by your queries.

On the unused indexes query, keep in mind this is reset if MySQL Server restarts. Some indexes are used infrequently, for example for monthly reports. So they could be identified as “unused” even though they are needed. One should keep notes in a wiki or something to document why each index was created, and identify the code and query that needs them.

I’d add to the list of schema design code smells the use of FLOAT or DOUBLE data type. Often it should be DECIMAL instead, especially for storing currency values.