Re: Faster distinct query?

I would look into pre-loading the lookup table (and pre-emptive 
maintenance).  Add the foreign key, but not the trigger.
>
> That makes sense. Thanks!
>
Yeah, then I got to wondering: Do you care?  Are these stations likely 
to be spoofed?  You have the station id and type in you data table and 
essentially the same in your lookup table.  If you're not replacing the 
id+type in your data table with a lookup id you really don't need to 
even have a foreign key.  Maybe sync them regularly but I'm not seeing 
the value in the runtime overhead.  Now presumably the station table is 
entirely pinned in memory and foreign key check might not be much 
overhead but it won't be zero.