Slow trigger in cluster-environment

I'm trying to keep a column up to date in NT-box from HP-UX using a trigger create trigger updaloupdate of al_vara on alotmnt referencing new as post for each row(update eestin@web_consy:alotmnt set al_vara = post.al_vara where al_serial=post.al_serial ) This is working reasonably well when the Unix side is running in one machine. Tests shows about 100 updates per second. When the Unix side is a cluster having two HP-9000 connected together same trigger works about 20 times slower !!! I've two cluster-configurations having same problem. Any ideas how to start tracking down the problem? Thanks Pentti