DSO to Cube loading is very slow

[Former Member]

Hello Experts,

I have one DSO which has more than 16 million entries and it took around 1 hour to load from PSA.

However when I am trying to load it to Cube from DSO it taking very long to finish.

I am loading fiscal year wise but then also for 2 million entries its taking very long time and not finishing.

Is there any solution for this?

This is first time this cube is getting loaded.

index has been dropped before loading.

there is no routine in the transformation.

Could anyone suggest something here?

Regards,

Ankie

[former_member185132 · Accepted solution]

Is the DSO indexed on Fiscal Year? Is Fiscal Year the first column of the key fields of the DSO? If the answer to both those questions is "No", then performance is likely to be slow. In this case it doesn't matter whether it's pulling 2 million records or 2000, it's doing a full table scan both ways and that takes time.

The other thing that can kill performance is Semantic grouping in the transformation/DTP. If there are any, and again if those are on non-indexed or non-key fields of the DSO they performance will be poor. Try to cut down on them if possible.

[Former Member · Accepted solution]

Hi Suhas,

There is no semantic group in DTP.

However fiscal year is not present in the key field section. its in data fields in DSO.

Regards,

Ankie

[KodandaPani_KV]

2.Hi,

1.increase the parllel process at the DTP .

select go to menu -> select setting for DTP batch manager -> increase the process.

2. decrease the data packet size -> 30,000 under extraction tab.

3. check the symantic group options.

4. before loading the data first delete the index and after loading create the index again.

Thanks,

Phani.

[Former Member]

Hello,

I have attached the file.

Please take a look.

all packages are doing the same.

Regards,

Ankie

[former_member185132]

The screenshot does not contain the time (duration) of each of the steps. As this is a discussion about performance, that information is vital.

So please post a screenshot that contains the duration column.

While you are doing that, look at the durations and identify the step within each packet that is taking the maximum time. Expand that in your screenshot so we can see which substep within that is actually taking time and analyze.

[karthik_vasudevan]

Hi Ankie

Could you please check if SIDs Generation upon Activation flag is unchecked in your source DSO.

If so, please check that and reload the data into the DSO. The try loading into your infocube and see the outcome.

Regards

Karthik

[Former Member]

Hi Suhas ,

I have attached new screenshot.

Regards,

Ankie

[Former Member]

Hi Ankie,

Could you please increase the number of process in the "Batch manager settings" in the DTP level and decrease the data package size.

Regards

Vivek

[Former Member]

Hi Vivek,

Default size has been put as 50000 for this DTP.

will it help if I increase the size of data packet?

We have 15 background process assigned to this DTP.

Regards,

Ankie

[Former Member]

Hi Ankie,

You can try that mean while how many years of data are you trying to extract in one load ?

Regards

Vivek

[karthik_vasudevan]

Hi Ankie

There is no need to increase the data packet size. It will make the performance worse in this case.

Could you please share the detail tab screen of your DTP monitor to see which step is taking more time.

Is there any sematic group selected?

Regards

Karthik

[Former Member]

Hi Vivek,

We have last 6 years data in DSO.

let me try that and will let you know the update.

Meanwhile if you came to know anything else please share with me.

Regards,

Ankie

[Former Member]

Hi Ankie,

As Karthik requested, Can you please send the screenshot of the DTP monitor so that we can identify which step is taking long to complete.

Regards

Vivek

[Former Member]

Hi Karthik,

There is no semantic group selected in the DTP.

Regards,

Ankie