Server RAM peaking on update statement within trigger

I've noticed server process RAM raising while executing simple update on table with no rows. From normal 170-350MB in RAM it peaks up to 2200 MB. I can decrease it back with "flush tables". The table has trigger on update. There is problematic update in it of such form:

UPDATE spac y,v_spac_updated u SET

 y.ph_id               =u.ph_id

,y.src_id              =u.src_id

...

where 

y.dp_id=u.dp_id AND y.sp_id=u.sp_id AND

decode(y.ph_id               ,u.ph_id                 ,0,1)

+decode(y.src_id              ,u.src_id                ,0,1)

>0

where v_spac_updated is a view.
While not from trigger this update presented executes fast and no RAM peak occurs, but from trigger it does.
Seems like some inefficiency in resource managing. Could this be improved?