complete refresh and network
Recently I encountered performance issue when I performed complete refresh for snapshots .
I did 3 test , the 1 st and 2 nd run much faster than 3 rd , though 1 st and 2 nd are DDL , they made less undo/redo than 3 rd , but it didn’t explain why the elapsed times were so different .
After investigating , I guess the cause was SQL*Net roundtrips to/from dblink .
Did you face such a issue ? Are there any parameters/approaches we can reduce the round trips between master and snapshot when doing complete fresh ? or please share me if you think the root cause is something else .