cancel
Showing results for 
Search instead for 
Did you mean: 

Table Cluster CDCLS in 4.6c Archive

Former Member
0 Kudos

Hello Members,

I am exporting/importing ECC 6.0 after upgrade from 4.6c.

I have an issue related to table CDCLS(table size 40GB, index 13GB), as per SAP recommandation, it should be archived or reorg before exporting for best performance.

Let me know facts before starting table archiving related to CDCLS.

Early response will be highly appreciated.

Regards,

Amit Lal

Accepted Solutions (1)

Accepted Solutions (1)

Former Member
0 Kudos

Hello Amit,

Unfortunately CDCLS is used in many many cases...

Call Tcode DB15 - select "Archiving Objects" - type CDCLS in "Objects for Table" - Enter.

You will get a list of archiving objects. The functional guy can tell which objects contain most of "changed" documents and whether they are appropriate for archiving.

Best regards,

Victor

Answers (2)

Answers (2)

Former Member
0 Kudos

Hi Amit,

I've had the same problem. My first export of CDCLS (65GB) took 72 hours (either your CPU is faster or you run a more recent R3load). After splitting the table into pieces, it run finally within 5 hours 12 minutes and it imports in 4 hours 50 minutes; running all other tables in parallel (not all of them at the same time - around 25 jobs at a time). Only for CDCLS I need 11 dual core IA64 (Montecito 1.6GHz) on HP-UX (export and import). It depends only on your I/O and your CPU how fast the export could be.

I think that the current SAP kernel contains the R3ta tool which could help you splitting the table, if you don't want to do it manually (which could be risky). Please read note 952514 - Using the table splitting feature.

Best regards

Ralph

Former Member
0 Kudos

Hello Ralph,

Thanks for your repsonse.

I checked with my team, we need to use split technology for CDCLS table.

We had done QA system import/export. Now further, preparing for Production, since we need minimum downtime, we have to use these tools.

I will highly appreciate if you can send more detailed steps about split method particularly for CDCLS with some screens, if available. Since we are new to use this method.

Thanks in advance,

Amit Lal

amitkumarlal@gmail.com

Former Member
0 Kudos

Hi Amit,

please read the note I posted before. It have several references to other notes which describes in detail how to do it. There is also a WIKI about Migration, you may read it too.

For a mail or a message here it is too complex. It depends on your data and on the available hardware. You may be also need to use the distribution monitor (note #855772) to distribute the load onto a set of application servers.

Best regards

Ralph

Former Member
0 Kudos

Hi Amit,

CDCLS will have change documents from many applications and so it will be covered by many archive objects. Please refer to the Data Management Guide for the options to deal with CDCLS. It also tells you how to use transaction TAANA to find out the applications which have the change documents in CDCLS. Find the top objects which is contributing to the growth of CDCLS and you should target those objects for archiving, so that it archives change documents along with it.

Hope this helps,

Naveen

Former Member
0 Kudos

Hello Naveen,

Now customer not want to archive the table for performance issue in archived logs.

Can REORG this table for better export timing...in Development box it tooks 30 hrs.

to export ...still going on...better way to optimize.

Thanks for your answer.

Amit lal

Former Member
0 Kudos

Hi Amit,

Reorg can help if you have a lot of fragmentation in the table. fragmentation will occur only in certain cases, for example when you archive and take the data out, else there is very less chance of fragmentation. so I dont think reorg on the table would help.

First you should do the table analysis to find out to which applications the data belong. If the major chunk of data is from change documents of master data, you can use archive object CHANGDOCU. If the data is from change documents of transaction data, then it would get archived from thier respective applications.

I am not much aware of the export, do you use index or is it direct table reads? you could try optimizing the index quality and updating the database statistics for cost based optimizer.

Hope this helps,

Naveen