I have installed Golden Gate in both of them and configured for bi-directional
replication.

I have created a table with primary key in both the servers –

create table EMPDATA (
empid number(10) primary key,
ename varchar2(30),
age number(2,0),
salary number(7,2)
);

I have inserted a record in .33 server as -
insert into EMPDATA values ( 100, ‘ABC’, 35, 25000 ); commit;
data got inserted in .33 and also got replicated in .89 server

I have inserted a record in .89 server as -
insert into EMPDATA values ( 101, ‘XYZ’, 49, 36000 ); commit;
data got inserted in .89 and also got replicated in .33 server

I have updated in .33 as -
update EMPDATA set salary = 25999 where empid = 100; commit;

Now my pump process in .33 server got slow and got a lag of 10 minutes
it got updated in .33 but DID NOT REPLICATED in .89 due to the lag

In .89 server I made an update -
update EMPDATA set salary = 30000 where empid = 100; commit;
It got updated in .89 and got instantly replicated in .33 server
So the salary for empid 100 in both .33 and 89 is currently 30000

Now as soon as the lag resolved to zero in pump process in .33 server
the salary in .89 changed to 25999 where as the salary in .33 server is 30000

Can this be prevented ? I want the latest update to be reflected in both
which should be 30000 in both .33 and .89 server

KIndly help

1. How many tables do you have in your database?
2. How often you restart the Extract process?

I get high consistent gets for

SELECT nvl(iot_name, ‘NULL’) FROM all_tables WHERE owner = :owner AND table_name = :object_name

Buffer gets ( in 30 mins): 199,569,171 blocks.
Executions (in 30 Mins) :134,267 blocks.
Gets/Exec (in 30 Mins) : 1,486.36

becoming the most top query. Replicated Table is only 1 table. And is not IOT but partitioned. How can I get rid of this high cost?

I have about 20 more tables to be defined for replication.

Thanks.

question on , golden gate extract process.

how do i fornce/ensure the extract process reads from the archvies only.

Thanks
NR