Celia, I had a long discussion with Glenn about the problems of the iita machine. We think the main problem is with the queue fragmentation because of the limited amount of real memory. Glenn sugests borrowing some memory from some other machine or division to see the results. He thinks 64-128 MGs might be enough to solve the problem. I think it might be worth a test. Please forward this message to Peter. RObb... =============================================================================== Robb Kambic Unidata Program Center Software Engineer III Univ. Corp for Atmospheric Research address@hidden WWW: http://www.unidata.ucar.edu/ ===============================================================================