Hi, Adv replication (version 9i, high usage OLTP) works fine most of the time and sometimes falls behind. Looks like it is doing different number of transactions/min at different times. There is no real difference in transaction size from time to time. Is there a way to know where the bottleneck is ? And more importantly how to improve it. So as to atleast have some consistency. Yes, I did the usual statspack, checking the tables etc. looking at metalink notes. Thanks