Table Splitting and Package Splitting

[Former Member]

Dear Gurus,

We have an ECC with 4.3 TB of database running on MSSQL 2008 and planning to migrate to SAP ASE.
We plan to use SWPM parallel export/import feature.

I understand that the table splitting using R3ta can be applied for either Top 20 or so biggest tables or tables with longer export runtime (based on the earlier mock runs).

But what I am unable to figure out is that what should be approach for splitting the table. E.g, we have CDCLS cluster table of around 300GB. and couple of more tables with 500GB, 400GB etc.

I have following queries regarding table and package splitting.

1. What should be min and max splitting ratio per table?
2. Does the behaviour of R3ta (R3ta runtime) increases with the more number of splits per table?
3. What aspects need to be considered before deciding on the splitting factor of a table?
4. Once the R3ta split is done, what should be ideal ranges for spltting?

Request you to guide me in this regard.

Thanks in advance.

Regards,
Ashok

[Reagan · Accepted solution]

As far as I know there is no min or max split ratio. If you start the table splitting preparation with more number of R3ta processes you should make sure there is sufficient CPU and memory available on the machine. You need to list (identify) the biggest tables in the system and decide how many splits you want to have.

Eg: A 20 GB table can be split into 4 packages which means each package will be of 5 GB in size.

[Former Member · Accepted solution]

Thanks Reagan.

I got answers for my 1st and 3rd questions.

2. Does the behaviour of R3ta (R3ta runtime) increases with the more number of splits per table?

4. Once the R3ta split is done, what should be ideal ranges for package spltting?

Can you please comment on the above queries? Thanks again.

Message was edited by: Ashok Varma

[Reagan · Accepted solution]

Hello Ashok,

When you run the table splitting preparation the R3ta tool is called and this will create the WHR files. If you run the preparation with just one R3ta then it is going to run for some hours, depending on the number of tables to be split. Instead, if you run multiple R3ta processes then the job will be done quickly. Again if the system does not have sufficient CPU and memory then there is no point in having multiple R3ta processes triggered. if the R3ta processes are more then the duration taken for the table splitting will be less (if the hardware resources are sufficient).

For the package split, last time I did the split I went ahead with the automatically populated values. Don't have the system to check and I don't remember it.

Regards

Reagan