Re: What is the best way to replicate few tables between 2 Oracle 10g severs ?
From: <stevedhoward_at_gmail.com>
Date: Fri, 17 Apr 2009 05:39:01 -0700 (PDT)
Message-ID: <fb22d1da-9b6d-4ca9-bde9-3e6627804517_at_g19g2000yql.googlegroups.com>
On Apr 16, 2:50 am, Veeru71 <m_ad..._at_hotmail.com> wrote:
> We need to replicate (1-way) a bunch of tables from Server-A to
> Server-B in close to real-time.
> Server-B is located in a foreign country.
>
> Apporx. # of tables to be replicated : 50 (size : 50 GB)
> Approx amount of data that gets replicated (changed data) per day :
> 100 MB
> Oarcle Versions : 10g, Solaris
>
> Constraints (due to some security reasons):
> - Server-B SHOULD NOT have any DB-Links to connect to Server-A
> - No UserId/Password of Server-A should be used anywhere in scripts/
> code/config files, etc on Server-B.
>
> What is the best way to achieve this ? Initially we thought of using
> Matreialized Views, but because of the above constraint, it may not
> be feasible.
>
> Do you think Oracle Streams would work ? Would it be an over-kill ?
> Any other options ?
>
> Thanks for your help.
Date: Fri, 17 Apr 2009 05:39:01 -0700 (PDT)
Message-ID: <fb22d1da-9b6d-4ca9-bde9-3e6627804517_at_g19g2000yql.googlegroups.com>
On Apr 16, 2:50 am, Veeru71 <m_ad..._at_hotmail.com> wrote:
> We need to replicate (1-way) a bunch of tables from Server-A to
> Server-B in close to real-time.
> Server-B is located in a foreign country.
>
> Apporx. # of tables to be replicated : 50 (size : 50 GB)
> Approx amount of data that gets replicated (changed data) per day :
> 100 MB
> Oarcle Versions : 10g, Solaris
>
> Constraints (due to some security reasons):
> - Server-B SHOULD NOT have any DB-Links to connect to Server-A
> - No UserId/Password of Server-A should be used anywhere in scripts/
> code/config files, etc on Server-B.
>
> What is the best way to achieve this ? Initially we thought of using
> Matreialized Views, but because of the above constraint, it may not
> be feasible.
>
> Do you think Oracle Streams would work ? Would it be an over-kill ?
> Any other options ?
>
> Thanks for your help.
For the amount of data you are talking about (100MB per day), you can probably get away with just triggers on the DB A tables that insert into DB B.
The streams option always bothers me, personally. We used it to migrate a fairly large database last year on what was then 10.2.0.3 from AIX single node to Linux RAC, and we had all kinds of weird issues, about 20 one-offs to apply, you name it. We were beginners with streams, so I'm sure we were inefficient in how we set it up, but I wouldn't call it "easy" to configure, especially with database that generate a lot of redo.
YMMV, Steve Received on Fri Apr 17 2009 - 07:39:01 CDT