Replication - Single Master DB

Question

Hi
I have a DB running on Server 1 and I want to create replication of all the Objects(existing Tables and any new table created) from Server-1 to Server-2.
I want One-Way replication only and it should be SYNC with SERVER 1, So Whats the best Replication approach ? Whats the performance impact. Please let me know if there is any tutorial available.
INFO
1. DB on Server 1 has 125 Tables.
2. The idea behind this that there are bunch of users who wants to query the DB , so we want to bring the same datbase running on SERVER 1 to another SERVER II , so that there want be any IO sharing between the PRODUCTION USER and QUERY USER.
Thanks
JR.

Answer 1

1. FAST REFRESH ON COMMIT cannot be done across a DBLink. I've explained this a few times in this forum. An ON COMMIT operation is a push operation : thus when the source table is updated by DML it has to push the updates to the MV. Since an MV across databases is via a DBLink that allows the TARGET to connect to the SOURCE and query the SOURCE, the SOURCE has no way to push the updates.
See http://hemantoracledba.blogspot.com/2008/06/mvs-with-refresh-on-commit-cannot-be.html
2. When you create an MV with a Refresh Interval, Oracle automatically submits a Job in the DBA_JOBS queue. This job does the actual refresh. Query USER_JOBS in the target (SERVER-2) database to check the job. Most likely, your SERVER-2 database has "job_queue_processes" set to 0 which disables running of jobs. You need to issue an
ALTER SYSTEM SET JOB_QUEUE_PROCESSES=1;in the target (SERVER-2) database so that it can run the Refresh job that will pull the updates.
(You could set job_queue_processes higher if you are going to run multiple refresh's in parallel).
On problem I foresee with your step to use MVs is that seperate MVs on individual tables will not synchronise data across the MVs.
If a transaction in the source database updates two tables "A" and "B" and the target database server uses two seperate refresh's for the two MVs "MV_A" and "MV_B" then "MV_A" and "MV_B" will not be synchronised to the same point in time (ie transaction).
You have to use Refresh Groups. MVs for all the the tables that could potentially participate in a transaction must be in the same Refresh Group and the Refresh job would be defined at the Refresh Group level . A problematic side effect of using Refresh Groups is that , in order to make the MVs transactionally consistent, Oracle must use DELETE and INSERT operations on all the MVs in the group. This will take time and generate significantly more UNDO and REDO in the Targer (SERVER-2) database.