Goldengate replication performance

Hi ,
This is about Goldengate replication performance.
Have configured Goldengate replication between OLTP and Reporting and the business peak hours occurs only for a one hour.
at that time I can see a LAG on the replicat side of around 10-15 minutes.
Rest of all the time there is no LAG.
I reviewed the AWR report of the target and I could see all the replicat process are executing with a elapsed time of 0.02 or 0.01 seconds.
However I could see a major wait event DB sequential read of 65%-71% of DB time and it is having the maximum Waits. apart from this there are no major wait event contributing to % DB time.(21% od DB CPU which I believe is normal )
and I can also see few queries are being hit at that peak time since it is a reporting server. and they are using select query which executes more than 15-20 minutes specially on the High transaction table.
Can you please advise where I should look on to resolve the LAG during the Peak hours.
I believe the select operation/wait event is causing the LAG during that peak hours. Am I correct.
can you please advise from your experience.
Thanks
Surendran

Hi Bobby,
Thanks for your response.
Please find my response as below,
Environment details as below.
1. Source and target DB - oracle database enterprise edition v 11.2.0.4
2. Goldengate version. 12.1.2.0.0 (Same Golden-gate version installed on source and target)
3 Classic CDC process is configured between source and target Goldengate.
Queries and response
Is there any long running transactions on the extract side?
No, long running transaction is seen, I can see a huge volume of transaction getting populated (over 0.3M records in 30 minutes)
Target environment information
High transaction DML activities is seen only on 3 tables.
As the target is reporting environment I can see many sql's querying those 3 high transaction populating tables and I can see DB sequential read wait event spiking up to 65%-71%.
I can also see in the AWR report that the GG session are executing the insert/update transaction is less than 0.01/2 sec.
Have to set the report for every 10 min. I will update to 1 min and share the report.
My query is : Is the select operation executed on that high transaction table on the reporting server during that high transaction window is causing the bottleneck and causes the LAG during that peak hours ?
or Do I need to look on other area's ?
Based on above information If you any further comments/advise please share.
Thanks
Surendran.

Similar Messages

How to reconfigure and restart GoldenGate Replication from MySQL to Oracle after a DDL operation on the source

I succesfully configured and performed the goldengate replication from MySQL 5.1 to Oracle XE 11g using GG 11g as per
http://www.oracle.com/webfolder/technetwork/tutorials/obe/fmw/goldengate/11g/GGS_Sect_Config_UX_MSQ_to_UX_ORA.pdf
After replication was successully started working I added a new column to one of the test tables TCUSTMER
ALTER TABLE TCUSTMER ADD COLUMN ADDRESS VARCHAR(64) DEFAULT NULL;
And inserted another row in the same table.
INSERT INTO TCUSTMER VALUES ('CLARK', 'CLARK BOATS', 'REDWOOD', 'CA', 'Test Addresss1');
As soon as I did that the replication broke.
2015-04-16 17:42:44 ERROR OGG-00146 VAM function VAMRead returned unexpected result: error 600 - VAM Client Report <CAUSE OF FAILURE : Table Metadata Altered cause
WHEN FAILED : While processing table map event in log processor
Then I reverted the DDL change back to original
ALTER TABLE TCUSTMER DROP COLUMN ADDRESS;
and tried to run the replication again
But it is not starting as before.
What should I do to move forward with same setup.

Hi ,
Whenever you do a DDL change in the Source table, you have to make the same changes in the Target side also. Then recreate the Definition File, copy it to the Target side and Start the GoldenGate processes.
1. Add the column in the Target side.
2. Recreate the Definition File and copy it to the Target side.
3. Recreate the Replicat process in the Target side and try to start it.
Hope this works..!!!!!!!! Any other suggestions friends????????
Regards,
Veera

How can we find if an oracle database is involved in GoldenGate replication

Just by connecting to an oracle database using sqlplus, is there anyway to know if a database is involved in GoldenGate replication?
i think dba_feature_usage_statistics is one option.
Thanks
Sarayu

Hi,
The best way that comes to mind is to make a grid where you have the entity and value dimensions on the rows and years and periods on the columns.
For entities you select the parent and the child you want to look at, value dimension members <Entity Currency> and [Proportion]. For years you select those which you want to look at and all the periods.
On the grid the [Proportion] should turn from green to red on one of the year/periods and that's the moment your entity became inactive to that parent.
Hope this helps.

GoldenGate Replication Polling Interval

I've been experimenting/evaluating GoldenGate replication and I keep running up against a limiting factor. If I understand things correctly, GoldenGate's extract processes monitors some of the logs Oracle spits out and uses them to generate its replication messages. However, as far as I can tell, those logs are read, at best, once a second (under heavy load, I've seen it take much longer). On the one hand, I expect there's still a great deal of tuning to be done, both on the hardware and networking side, but on the other, it severely concerns me that there's a baseline minimum one second delay for each replication hop due to a polling period from GoldenGate alone.
So the real question is: can the GoldenGate processes' polling period can be reconfigured to be quicker? I couldn't find anything in the documentation to suggest this was possible, but I'm convinced it should be.

We tried using the variables, but the results weren't what we expected. We're running Goldengate on Amazon EC2 Oracle instances, so we suspect there may either be misconfiguration of Oracle or amazon's network being noisy to confuse things.
Also the project is currently on hold, so I can't investigate it further. However I'm content that the question has be answered sufficiently, it's just a shame we couldn't investigate them further.

GoldenGate replication of creating users, roles

System Specs:
O/S : RHEL 5 (Tikanga)
RDBMS: 11.2.0.3 (Standalone, ASM, Archivelog)
GoldenGate v:11.1.1.0 (getting ready to upgrade to v11.1.1.2 (want to utilize the ADD SCHEMATRANDATA and updated sequence support)
I have read the documentation in GG that says that DDL support Oracle restricted schemas is not supported (including sys and system). From this document a create statement is considered DDL.
However, I have to think that when a user or role is created on the Source that you want that action replicated to the Targets? So you don't have to rerun the action on the number of targets you have. This is the benefit of replication, correct?
Without this ability replication is somewhat restrictive.
Please someone shed some light on how a user/role statement could be replicated?
Thanks!
Jason

Okay, so I reviewed the documentation and need some help in replicating DDL for 2 schemas.
Here is my Extract and Replicat modules
EXTRACT EXT1
USERID ggs_owner, PASSWORD ggs_owner
RMTHOST db2, MGRPORT 7809
RMTTRAIL /home/oracle/goldengate/dirdat/gg
---ddl---
DDL INCLUDE MAPPED
---dml---
table schema1.*;
table exclude schema1.*_sq;
REPLICAT REP1
ASSUMETARGETDEFS
USERID ggs_owner, PASSWORD ggs_owner
DDL INCLUDE MAPPED
DDLERROR DEFAULT IGNORE RETRYOP
REPERROR (1403, DISCARD)
MAP schema1.*, TARGET schema1.* ;
MAP schema2.*, TARGET schema2.*;
My issue is that ddl replication is only working for schema1. No ddl is being replicated with schema2. I know there must be a something small I am overlooking.
I was successful with
DDL INCLUDE ALL, EXCLUDE "schema3.*, EXCLUDE "schema4.*", EXCLUDE "schema5.*", etc...
However, I don't want to list out every schema that could potentially perform ddl. I would think the DDL INCLUDE MAPPED and then include those schemas that you want mapped.
Any ideas?

Goldengate replication of CLOBs in Linux

Hi All,
I'm facing a weird problem using GoldenGate to replicate CLOBs in a Linux environment. I have configured GG for a simple one-way replication of no more than 10 tables from Oracle to Oracle databases.
Most of these tables have CLOB columns and during my testing, in a Linux box, I detected that the CLOB data is replicated to the target db adding an extra space between each character. For example:
In the source:
Text (CLOB): Testing replication of CLOB using Goldengate
In the target, the replicat is saving the data in this way:
Text (CLOB): T e s t i n g r e p l i c a t i o n o f C L O B u s i n g G o l d e n g a t e
Have you seen this problen before? Is there any problem with my GG installation? I tested this in a Windows platform and there, the CLOBs are being replicated fine.
The details of my Linux and GG env:
SO: Oracle's Enterprise Linux 32bit
Database: 11GR2
GG (I’ve tried both 11G and 10G versions):
For 10G:
[orgolden@naciprdell209 golden]$ ./ggsci
Oracle GoldenGate Command Interpreter for Oracle
Version 10.4.0.19 Build 002
Linux, x86, 32bit (optimized), Oracle 10 on Sep 17 2009 23:49:42
Copyright (C) 1995, 2009, Oracle and/or its affiliates. All rights reserved.
For 11G:
[orgolden@naciprdell209 golden_11g]$ ./ggsci
Oracle GoldenGate Command Interpreter for Oracle
Version 10.4.0.19 Build 002
Linux, x86, 32bit (optimized), Oracle 11 on Sep 17 2009 23:49:42
Copyright (C) 1995, 2009, Oracle and/or its affiliates. All rights reserved.
Thanks and regards,
Harvey

Okay, so I reviewed the documentation and need some help in replicating DDL for 2 schemas.
Here is my Extract and Replicat modules
EXTRACT EXT1
USERID ggs_owner, PASSWORD ggs_owner
RMTHOST db2, MGRPORT 7809
RMTTRAIL /home/oracle/goldengate/dirdat/gg
---ddl---
DDL INCLUDE MAPPED
---dml---
table schema1.*;
table exclude schema1.*_sq;
REPLICAT REP1
ASSUMETARGETDEFS
USERID ggs_owner, PASSWORD ggs_owner
DDL INCLUDE MAPPED
DDLERROR DEFAULT IGNORE RETRYOP
REPERROR (1403, DISCARD)
MAP schema1.*, TARGET schema1.* ;
MAP schema2.*, TARGET schema2.*;
My issue is that ddl replication is only working for schema1. No ddl is being replicated with schema2. I know there must be a something small I am overlooking.
I was successful with
DDL INCLUDE ALL, EXCLUDE "schema3.*, EXCLUDE "schema4.*", EXCLUDE "schema5.*", etc...
However, I don't want to list out every schema that could potentially perform ddl. I would think the DDL INCLUDE MAPPED and then include those schemas that you want mapped.
Any ideas?

OCI Error ORA-01403: no data found in oracle goldengate replication after tts instaniation

I recently migrated our tg core system from sun solaris(11.1.0.7) to linux (11.2.0.3) environment using goldengate method (transportable tablespace method used for instantiation).
The initial replication worked with HANDLECOLLISIONS and after I monitor the lag finished, I took Handlecollisions off and noticed discarded report with OCI Error ORA-01403: no data found in two replications.
I followed every step in the tts migration steps provided by oracle best practice.
Can anybody provide any clue how do I fix this issue?
Thank you in advance.

extract and replicat are on schema level.
DO I have to do anything for replication on schema level?
Basic trandata logging enabled on source.

GoldenGate Replication or Oracle 11g Streams?

We are working with a customer that is interested in implementing Oracle 11g Streams.
I don’t know the GoldenGate products and
I am not real familiar with streams other then hearing it can be tough to tune and if something goes wrong, it can be a pain to get right..
They recently heard of Oracle’s purchase or GoldenGate and became interested when they read that the company may offer possibly a simpler replication strategy then streams as far as setup, tuning, maintainin, etc..
Has anyone heard of what Oracle’s future plans/direction are for integrating GoldenGate into their fold, specifically the replication side like streams?
Thanks

I predict that in the future the GoldenGate Oracle users are going to get the short end of the stick now that Oracle has purchased them. I do not believe that Oracle purchased GoldenGate for their Oracle to Oracle capabilities. I assume that they wanted to acquire their ability to replicate from non-Oracle databases.
There is nothing inherently wrong (design-wise) that makes the Oracle Streams product inferior to GoldenGate or Quest's Shareplex product. But both companies have continued to sell their products (which are not cheap) against a product which basically free.
Why is this? Do you think that Oracle does not have the technical capabilities to build a great replication solution? If Oracle was interested in making a great Oracle to Oracle replication product, they could have invested a small fraction of the amount they paid for GoldenGate and added the necessary resources to build the best replication product on the market. But the don't because that is not what they are interested in doing.
There are a couple of press releases that Oracle put out about the GoldenGate acquisition and one of the advantages they claim GoldenGate will benefit from is their $30 Billion dollar R&D budget. But why didn't Oracle use some of that 30 billion to make Streams a great product??
Oracle purchased GG for reasons that have nothing to do with Oracle to Oracle replication. Future resources and priorities will go towards those goals.

GoldenGate Replication 32bit db to 64bit db

Hi all,
I have two (2) 9i databases (9.2.0.6) on OEL 5.3(32/64b). One is 32bit, and the other is 64bit.
Can I replicate 32bit database going to 64bit?
Thanks a lot for your unfailing support.
mk

Hi MK,
Another question is, how many numbers tables limitation does GG has? it it 2,000 tables maximum only?
GoldenGate Extract Process does not have limitation on number of Tables to be replicated. It all depends on the Memory. Oracle GoldenGate supports 5000 concurrent Extracts and Replicat Processes per Instance of Oracle GoldenGate. Each Extract and Replicat process needs approximately 25-55 MB of memory, or it can also be more depending on the size of the transactions and the number of concurrent transactions. This is for Classic Capture and Replicat mode. We have also another mode called Integrated Capture and Integrated Replicat mode. For setting Memory when using Integrated Capture Mode use the below.,
http://www.oracle-scn.com/?p=195
Another question is, can I replicate at database level ? so I won't need to type each table 1-by-1?
Database Level replication is not possible in Oracle GoldenGate. Only Data level Replication is posssible. You can use the wildcard ' * ' . For example, If you need to replicat a schema SCOTT, You don't need write separate entries for each tables instead you can use the wildcard(*). Please see the below example,
TABLE SCOTT.TABLE1;
TABLE SCOTT.TABLE2;
TABLE SCOTT.TABLE3;
Instead of using the above, you can use the Wildcard entry.
TABLE SCOTT.*;
The above automatically picks up all the existing tables and also the tables which are newly added.
Is it supported also to replicate tables from 9i 32bit Db to 11g 64bit Db?
Yes, it is supported.
How can the two get in sync if they have different version of GG installed?
It is recommended to use the same version on both the source and target. However, If the Oracle GoldenGate versions on the source and target database are different, it is a good practice to use the RELEASE option of the EXTFILE/RMTFILE and EXTTRAIL/RMTTRAIL parameter, especially if you have a lower version of Oracle GoldenGate on the target.
Example: EXTFILE ./dirdat/em, FORMAT RELEASE 10.4
Regards,
Veera

GoldenGate Replication - Between Schemas On Same Host

Guys - My requirement is fairly simple.I have two schemas, GG [Source] and GGR [Target] on the same host. Have one table called GG.SYNC_TABLE. I am having difficulties to push data from GG to GGR
Below are the extract and replicat information
EXTRACT EXT_AP1
SETENV (ORACLE_SID=ERPA4)
RMTHOST mdeagaix825, mgrport 7809
USERID GG@ERPA4, PASSWORD goldengate1
DISCARDFILE ./dirrpt/ext_ap1_discard.rpt, append, megabytes 50
RMTTRAIL ./dirdata/sa
TABLE GG.AP_AE_HEADERS_ALL;
TABLE GG.AP_AE_LINES_ALL;
TABLE GG.AP_BANK_ACCOUNTS_ALL;
TABLE GG.AP_BANK_BRANCHES;
TABLE GG.AP_CARDS_ALL;
TABLE GG.AP_CHECKS_ALL;
TABLE GG.AP_CREDIT_CARD_TRXNS_ALL;
TABLE GG.AP_EXPENSE_REPORTS_ALL;
TABLE GG.AP_EXPENSE_REPORT_HEADERS_ALL;
TABLE GG.AP_EXPENSE_REPORT_LINES_ALL;
TABLE GG.AP_EXPENSE_REPORT_PARAMS_ALL;
TABLE GG.AP_EXP_REPORT_DISTS_ALL;
TABLE GG.AP_HOLDS_ALL;
TABLE GG.AP_HOLD_CODES;
TABLE GG.AP_INVOICES_ALL;
TABLE GG.AP_INVOICE_DISTRIBUTIONS_ALL;
TABLE GG.AP_INVOICE_LINES_ALL;
TABLE GG.AP_INVOICE_PAYMENTS_ALL;
TABLE GG.AP_NOTES;
TABLE GG.AP_PAYMENT_HISTORY_ALL;
TABLE GG.AP_PAYMENT_HIST_DISTS;
TABLE GG.AP_PAYMENT_SCHEDULES_ALL;
TABLE GG.AP_POL_VIOLATIONS_ALL;
TABLE GG.AP_SELF_ASSESSED_TAX_DIST_ALL;
TABLE GG.AP_SUPPLIERS;
TABLE GG.AP_SUPPLIER_SITES_ALL;
TABLE GG.AP_SYSTEM_PARAMETERS_ALL;
TABLE GG.AP_TERMS_LINES;
TABLE GG.AP_TOLERANCE_TEMPLATES;
TABLE GG.SYNC_TABLE;
REPLICAT REP_AP1
SETENV (ORACLE_SID=ERPA4)
USERID GG@ERPA4, PASSWORD goldengate1
ASSUMETARGETDEFS
REPORTCOUNT EVERY 1 MINUTES, RATE
DISCARDFILE ./dirrpt/rep_ap1.dsc, PURGE
MAP GG.AP_AE_HEADERS_ALL, TARGET GGR.AP_AE_HEADERS_ALL;
MAP GG.AP_AE_LINES_ALL, TARGET GGR.AP_AE_LINES_ALL;
MAP GG.AP_BANK_ACCOUNTS_ALL, TARGET GGR.AP_BANK_ACCOUNTS_ALL;
MAP GG.AP_BANK_BRANCHES, TARGET GGR.AP_BANK_BRANCHES;
MAP GG.AP_CARDS_ALL, TARGET GGR.AP_CARDS_ALL;
MAP GG.AP_CHECKS_ALL, TARGET GGR.AP_CHECKS_ALL;
MAP GG.AP_CREDIT_CARD_TRXNS_ALL, TARGET GGR.AP_CREDIT_CARD_TRXNS_ALL;
MAP GG.AP_EXPENSE_REPORTS_ALL, TARGET GGR.AP_EXPENSE_REPORTS_ALL;
MAP GG.AP_EXPENSE_REPORT_HEADERS_ALL, TARGET GGR.AP_EXPENSE_REPORT_HEADERS_ALL;
MAP GG.AP_EXPENSE_REPORT_LINES_ALL, TARGET GGR.AP_EXPENSE_REPORT_LINES_ALL;
MAP GG.AP_EXPENSE_REPORT_PARAMS_ALL, TARGET GGR.AP_EXPENSE_REPORT_PARAMS_ALL;
MAP GG.AP_EXP_REPORT_DISTS_ALL, TARGET GGR.AP_EXP_REPORT_DISTS_ALL;
MAP GG.AP_HOLDS_ALL, TARGET GGR.AP_HOLDS_ALL;
MAP GG.AP_HOLD_CODES, TARGET GGR.AP_HOLD_CODES;
MAP GG.AP_INVOICES_ALL, TARGET GGR.AP_INVOICES_ALL;
MAP GG.AP_INVOICE_DISTRIBUTIONS_ALL, TARGET GGR.AP_INVOICE_DISTRIBUTIONS_ALL;
MAP GG.AP_INVOICE_LINES_ALL, TARGET GGR.AP_INVOICE_LINES_ALL;
MAP GG.AP_INVOICE_PAYMENTS_ALL, TARGET GGR.AP_INVOICE_PAYMENTS_ALL;
MAP GG.AP_NOTES, TARGET GGR.AP_NOTES;
MAP GG.AP_PAYMENT_HISTORY_ALL, TARGET GGR.AP_PAYMENT_HISTORY_ALL;
MAP GG.AP_PAYMENT_HIST_DISTS, TARGET GGR.AP_PAYMENT_HIST_DISTS;
MAP GG.AP_PAYMENT_SCHEDULES_ALL, TARGET GGR.AP_PAYMENT_SCHEDULES_ALL;
MAP GG.AP_POL_VIOLATIONS_ALL, TARGET GGR.AP_POL_VIOLATIONS_ALL;
MAP GG.AP_SELF_ASSESSED_TAX_DIST_ALL, TARGET GGR.AP_SELF_ASSESSED_TAX_DIST_ALL;
MAP GG.AP_SUPPLIERS, TARGET GGR.AP_SUPPLIERS;
MAP GG.AP_SUPPLIER_SITES_ALL, TARGET GGR.AP_SUPPLIER_SITES_ALL;
MAP GG.AP_SYSTEM_PARAMETERS_ALL, TARGET GGR.AP_SYSTEM_PARAMETERS_ALL;
MAP GG.AP_TERMS_LINES, TARGET GGR.AP_TERMS_LINES;
MAP GG.AP_TOLERANCE_TEMPLATES, TARGET GGR.AP_TOLERANCE_TEMPLATES;
MAP GG.SYNC_TABLE, TARGET GGR.SYNC_TABLE;
Extract, Replicat and Manager processes are running fine. But a commit on target is not propagating the data across to GGR schema. Supplemental logging is enabled. Archiving is not [I hope its not required]. What do you think I am missing here?
I am fairly new to GoldenGate, Hence if you want me to run any commands, Please provide the commands. Thanks much in advance.

Duplicate post.

Goldengate Replication setup for windows and SQL 2000

HI
can anyone provide documentation on how to setup and configure Replication on windows platform
using sql 2000.
help is appriciated.
Regards
Manish

Manish ,
you can get all the steps from install guide .Hope you might have downloaded along with the product.Its pretty straight forward.
Thanks,
Antony Joseph Jethu

Goldengate Replication without software on the target

Hi
for a PoC I try to find out if there is a way to use Goldengate for reporting offloading without installing the software on the target. Does that work? I tried with a similar approach as with target on same host as source but get the error message ogg-00644 TNS: no listener.
Any hints are welcome

That's excatly what I did and where I received the above error message
Here are the parameters for extract and replicat:
EXTRACT EXTLOCAL
SETENV(NLS_LANG=AMERICAN_AMERICA.AL32UTF8)
USERID GGATE, PASSWORD xxx
DISCARDFILE /export/home/ggate/goldengate/dirdis/ext1.dsc, PURGE
RMTHOST localhost.com, mgrport 7809
RMTTRAIL /export/home/ggate/goldengate/dirdat/re
TABLE DEMO.GGTEST;
TABLE DEMO.GGMULTI1;
TABLE DEMO.GGMULTI2;
REPLICAT replocal
USERID [email protected], PASSWORD ggpwd
SETENV(TNS_ADMIN=/home/ggate)
SETENV(NLS_LANG=AMERICAN_AMERICA.WE8ISO8859P1)
SETENV(ORACLE_HOME=/data/oracle/product/10.2.0.5)
DISCARDFILE /home/ggate/goldengate/dirdis/repl1.dsc, PURGE
ASSUMETARGETDEFS
MAP DEMO.*, TARGET DEMO.*;
Edited by: carajandb on 06.02.2013 16:41

Goldengate replication with missing trailing blocks error

Hi Experts
I set up a simple goldengate test blob datatype at 10.2.0.4 in 2003 32 bit window server
It does not work.
in source emp_ext.rpt, I got an error message as
2010-06-30 15:53:13.701 Redo thread 1: Online log D:\ORADATA\SALE\REDO01.LOG on sequence# 159 has missing trailing blocks.
What means is about this message? how to fix it?
Thanks
Jim

"missing trailing blocks" refers to the fact that the blocks belonging to the redo.log thread that GoldenGate was extracting have been interrupted by a block not belonging to the thread in the log. This is usually just a warning message and indicates that GG will probably begin reading from the archive logs. Of course, if the archive logs are not still available, this could cause an ABEND.

GoldenGate + Auditing = Performance issue ?

Hello people,
we are running GG on Solaris for the replication part. Since we set AUDIT TRAIL = TRUE we had an increase from 5% to 30% in CPU.
Is it possible that the combination of GG + Auditing is responsible for this increase? Please note that I just change the parameter. No auditing on objects enabled.
Please for your opinions.
Thank you in advance.
Edited by: drbiloukos on 26 Ιουν 2012 3:13 μμ

drbiloukos wrote:
CPU increased immediately after seting AUDIT TRAIL and remained as such.
I do not say that the combination of GG and Audit is definetely the one to blame. I just need to know if there is an known issue, and your opinion on my suspicion.
It might be just an effect of Auditing, although Oracle claims aprox 5% of CPU overhead.
I will disable Auditing on first chance.The CPU change depends on what is being audited, and to where it is being audited (database, os file). Afaik, setting it to "true" isn't a valid value? Don't you have to specify what is to be audited (e.g., db vs. extended) and to where? (e.g., db, os file, xml, ...) By default it could have already been enabled (audit_trail=db) -- depending on your version of oracle & how the db was created (e.g, dbca). I'm not sure what the result of setting it to "true" would be...?
In theory, if enabling auditing on the target DB, then "replicat" -- as just a normal SQL client -- will increase load on the target when the volume of replication is high (just like any other sql client). If enabling auditing on the source, you have the exact same issue as the target (additional sql clients increase load) -- plus, if your audit info is written to the database, at the very least, "extract" has to parse through and ignore the audit info from the logs. (Worst case would be if your "extract" may inadvertently be picking up these records - but I don't think that will happen if logged to a "sys" table.)

Can we implement goldengate replication between Oracle to Oracle using RMAN

Dear All,
I have already implemented Oracle Golden Gate Between Oracle Database 11gR2 on Linux.
The method I have been using to copy database on Target is like this:
select dbms_flashback.get_system_change_number() from dual;
expdp user/password directory=backup_dir flashback_scn=1355907575 dumpfile=gg.dmp logfile=gg.log schemas=radius
scp radius_dsl_gg.dmp [email protected]:/backup
impdp user/password directory=backup_dir dumpfile=gg.dmp logfile=gg.log schemas=radius
I want to replace this data pump with RMAN, because most of the time I don't have enough disk space to take data pump in production server. How can I replace it with RMAN and how can i track new records inserted in production database like we use flashback_scn in data pumps. Or is there any other better solution.
Has anyone done this, kindly share your knowledge.
Regards, Imran

You don't need a dump file for DataPump, you can load directly through a database link thus eliminating the space required for the intermediate dump file.
With RMAN you will either be looking for a point-time DB restore/recover as of specific SCN and then you can start the Replicat from that SCN. You can do a TSPITR + TTS if replicated data is a small subset of the entire database. If you can put the entire tablsapce to be read only then you can get away with only TTS as well. You configure Extract before you start any restore/recover and that's how you track outstanding changes.

Goldengate replication performance

Similar Messages

Maybe you are looking for