Installation of Oracle GoldenGate on RedHat Linux AS 4

I am going to install and configure standard GoldenGate configuration in which there is on Extract Group on the source machine, which sends the data to be replicated to the trail on the target system through normal network. On target system, the Replicat process consumes the data from the trail.

I am using 32-bit Oracle 10gR2 on both source and the target systems and the operating system is 32-bit Redhat Linux AS 4. I have created sourcedb on sourcemachine and targetdb on targetmachine. I have made sure that I can access both databases from both machines through SQLPLUS.

The Oracle GoldenGate was downloaded from here.

Then a folden 'ggs' was created in ORACLE_HOME and it is the directory where the GoldenGate will be installed, I added the path '$ORACLE_HOME/ggs' to the shared library enviornment variable LD_LIBRARY_PATH.

I opened up the terminal window, made sure that the ORACLE_HOME, ORACLE_SID, and LD_LIBRARY_PATH variables are set and then copied the downloaded the zip GoldenGate software file to ORACLE_HOME/ggs and unzipped it there.

It produced a TAR file. I untarred it by doing:

$ tar -xvof

Then I changed directories to the GoldenGate directory and started GoldenGate by doing:

$ ggsci

GGSCI> CREATE SUBDIRS

GGSCI> START MANAGER

GGSCI> EDIT PARAMS MGR

It will open a parameter file for the manager process in the vi editor, insert following line in the file and save and quit:

PORT 7809

:wq

GGSCI> STOP MANAGER

GGSCI> START MANAGER

I installed Oracle GoldenGate and configured the Manager process on both source and the target systems.

Now I will create Extract Group on the Source system. I am using the SCOTT SCHEMA and replicating the table 'dept':

ADD EXTRACT ext, TRANLOG, BEGIN

Then specify the remote trail on the host:

ADD RMTTRAIL rmt, EXTRACT ext

Now create the parameter file for the EXTRACT Group by doing:

GGSCI> EDIT PARAM ext

It will open up a vi editor:

EXTRACT ext
sourcedb, scott, tiger
RMTHOST targetmachine, MGRPORT 7809
RMTTRAIL rmt
TABLE scott.dept;

This completes the configuration at the source system.

Now move on to the target system.

Make sure to install GoldenGate on target machine and configure the MANAGER process as was done on the source machine.

First create a checkpoint table on target system. I am creating it in the scott schema wit name'CSNTAB;.

GGSCI> EDIT PARAMS ./GLOBALS

It will open up a vi editor. Enter:

CHECKPOINTTABLE scott.CSNTAB

save and quit.

GGSCI> DBLOGIN targetdb, scott, tiger

GGSCI> ADD CHECKPOINTTABLE SCOTT.CSNTAB

Then create a REPLICAT GROUP:

GGSCI> ADD REPLICAT rep, EXTTRAIL rmt, BEGIN

Then create a parameter file for Replicat:

GGSCI> EDIT PARAM rep

It will open up a vi editor, Enter:

REPLICAT rep
SOURCEDEFS ASSUMETARGETDEFS
TARGETDB targetdb, scott, tiger
MAP scott.dept, TARGET scott.dept;

Enter and commit a row in dept table of sourcedb and then verify that it gets replicated to the targetdb's dept table of scott schema.

Why I am Interested in Oracle GoldenGate

Without any marketing hype and without any sales talk, I am already hooked to the Oracle GoldenGate. In this age, when the prices of gold are breaking all the records, it's so enriching to have a all-gold gate.

When you have 10 SQLSERVER databases, 20 MYSQL Databases, and 30 Oracle OLTP databases and 5 Oracle Datawarehouses and many XML data sources, and a small of team of DBAs have to manage the data flow between these disparate data sources.

In this sort of environment, challenges are simply gigantic. Integrity of information, freshness across the databases, agile response, data mappings, data volatility, Information lifecycle management makes information flow in any direction a titanic task.

It was not so easy to manage such kind of environ through naive replication and even Streams were a hassle. With Oracle Golden Gate, all the above challenges seem to be addressed out of the box, and they say that this is a mature technology with proven track record.

Can't wait to try it out, and then have it at my place.

Log Buffer

When Pythian blogs, it stays blogged. Pakistan's First Oracle Blog prouds to host the edition of Log Buffer.

Let's Start with Oracle Blogs:

Nuno Souto, the celebrated Kangroo Blogger comes hard at someone from Oracle, who blogged in not-so-gracious manner about the Oracle customers and those who don't like the new My Oracle Support.

Hey have it while its hot and sizzling. Angelo Santagata has announced that Fusion Middleware 11.1.1.2.0 Is available is now available to download.

Learn how to migrate and create uses by using DBMS_METADATA with the help of Alex Nuijten.

Here is another perspective on the EU's opposition to the Oracle's Sun acquisition.

Want to see the video demonstration of Sun Oracle Database Machine through Kevin's blog? Click here.

Nothing by Jonathan Lewis should be missed. How could one miss when he talks about Index Freelists?

Tim Dexter discusses the template problem in Siebel CRM integration.

Now Comes MySQL:

Mark tells about InnoDB Plugin which was added upon users' request.

Sahan Chaliha dishes out a fine introductory note on RDS.

If you are still wondering about the FLIFO Scheduling for the InnoDB, you may like to read this post by Mark Callaghan.

Want to refresh your MYSQL Tool set, take a look at this blog post by Gerry Narvaja.

Jeremy Zawodny does it again and enlighten us about the issues related to MySQL Performance.

Bye Bye Metalink

It seems that a friend expired. Someone waving out of the window of leaving train for an unknown destination, for never to return, as the Oracle Metalink retires this weekend.

My Oracle Support portal will be the official portal for Oracle's paid support. Following the upgrade, My Oracle Support will become the single online support platform for all Oracle products, and Classic MetaLink will retire.

For a long time, people will keep calling My Oracle Support as MetaLink. Will they also change metalink.oracle.com to myoraclesupport.oracle.com? Seems unlikely.

The new My Oracle Support is cool, flashy and satisfies Web 2.0. But good old metalink will be missed dearly.

How to Enable Tracing for Oracle Report Server

Report Server is the most toughest part in Oracle Application Server to manage, in terms of the load for concurrent users. It's configuration is pretty much straight forward, but when errors in report appear, and you bang your head and keep restarting the report server in vain, then tracing might help you to locate the error, or enable Oracle Support to get more in-depth information:

Edit rwservlet.properties from ORACLE_HOME/reports/conf folder as below
from
#TRACEOPTS=TRACE_ALL
#TRACEFILE=rwservlet.trc
#TRACEMODE=TRACE_REPLACE
to
TRACEOPTS=TRACE_ALL
TRACEFILE=rwservlet.trc
TRACEMODE=TRACE_REPLACE

5.Edit .conf file in ORACLE_HOME/reports/conf folder as below
from

to
< trace traceOpts="trace_all" traceMode="trace_append"/>

6.Restart the server,reproduce the problem and inspect all the files from ORACLE_HOME/reports/logs folder.