Hi,
We use Materialized Views to implement replication. A
is the center RAC database, several B are A's read-only Materialized
Views sites. Some Materialized Views are very big, so we need use Offline
Instantiation to create them or complete refresh. All A and B are very critical
7*24 system. Any recreation of Materialized Views is a disaster.
Now, for network limit, new sites can't connect to A
directly. So I design C to be B's read-only Materialized Views site, and
then sync with A indirectly.
My questions are:
1. cascading replication cause weak replication or
not? In past years, a B disconnect with A, and then A's Materialized
View log become so huge that B can't do fast refresh. The result is we
recreate mviews in B, take downtime on the master
table and truncate mview log in A. I worry about in
cascading replication one site network failure will cause more fix
work.
2. If A has a data guard database G, when A
switchover to G with no-data loss, whether it will affect B's mviews refresh.
Does any one have this kind of experience? I know Oracle use internal trigger
for mview, so I worry about whether this objects will be applied in
G.
3. Application sometimes need to modify the master
table in A. If I don't want to recreate mviews or change mview
to advanced replication, Is there any internal way to modify mview's trigger,
etc. to propagate DDL changes, like DBMS_REPCAT.EXECUTE_DDL procedure
does?
Best
Regards,
Tina
