Question

For a web application where deleting records happens often, what is the best, if we take performance into consideration?

• check if the record has any reletaionship with other tables and prevent the user from deleting.

  ex : the query will be like this

   if not exist(select * from table1 where tableX_id = @id) and 
         not exist(select * from table2 where tableX_id = @id)
         ...
      then 
         delete from tableX where id = @id 
  

  or

• perform the delete and let the RDBMS rais an error due to the foriegn key constraint

    try{
         Service.DeleteRecord(id)
    }catch{
         Handle the error here
    }
  

  in this case the query will be simple

   delete from TableX where id = @id
  

Answer 1

If the goal is to prevent rows from being deleted only in the case where actually running the delete would yield an error because it is referenced by other tables, then I suggest you check for the violation before letting SQL Server check for you. I’ve done some testing on this and letting SQL catch the error or just using TRY/CATCH can be more expensive if you expect even a moderate amount of failures. With the proper indexes in place, the additional cost of performing the check is negligible; however the cost of NOT doing the check is certainly LESS negligible.