Originally Posted by marcopo
the performance of UPDATE should not be the best.
A vague statement like "should not be the best
" is not
something you should use as the base for your decisions on how you implement something.
Implement your required functionality and only if you have performance problems in your
workload, take action to improve that.
An UPDATE using a primary key will be extremely fast, even when the table has several millions of rows.
Premature optimization - especially if it isn't based on hard facts and your own testing - is seldom a good idea.