What caused a huge increase in i/o in my Oracle database?

Hi,

  Does anyone have any ideas as to how I could identify a database session(s) which suddenly started to perform huge amounts of additional i/o? The reason for the question is that the database in a production RAC-one-node configuration relocated itself onto the 'other' instance recently and the clusterware evicted the node on which the original active instance was running because it became unresponsive. Subsequent examination of the 'vmstat' i/o report just before the time of the failure showed a huge, sudden increase in i/o between 08:21:31 and 08:22:01 (please see the attached extract for further details). The database instance, of