Running a trigger always has some overhead - at least you are doing a context shift from the SQL engine to the PL / SQL engine for each row that triggers the trigger. Although the absolute amount of overhead when starting a trigger is relatively constant, the percentage of overhead varies greatly depending on how you do the DML. If you have an application that adds or modifies rows in sets, which is the fastest way to work with relational data, triggers have a much greater relative impact on performance because the cost of these context shifts plus the cost of what the trigger is actually doing, quickly dominates the cost of running DML.
, . , , , , , , , , , , . .