it was a trick question. It is impossible to focus in on the I/O latency of specific background processes like LGWR, DBWR much less aggregate their specific I/O characteristics across nodes of a cluster... impossible, that is, without mxodmstat(8) as I describe starting on page 13 of this paper: http://www.polyserve.com/pdf/VMDB_BC.pdf reading only the ODM section of that paper makes people see just what they are missing yes..it is a shameless plug, but it is a technology solution to what a lot of people see as a problem in RAC environments..it is a single database, but you monitor the instances as separate entities by and large. We focus on the ability to both aggregate fine-grained information like LGWR only or PQO only, as well as drill down... Helps when you have a libarary that Oracle links with :-) -- //www.freelists.org/webpage/oracle-l