Optimizing Data Load for Large Replication Flow Object

[wsamuel]

Good day All,

I have a quick question regarding the optimization of data loading for a replication flow object containing over 400 million records. The object in question is the CDS view I_GLACCOUNTLINEITEMRAWDATA, which includes 348 columns.

We are using replication flows to transfer data from an on-premise S/4HANA system to Datasphere, utilizing an initial + delta load method.

I started the flow on Wednesday, but it stalled and ultimately failed on Thursday night after loading 170 million records out of the expected 400 million. The flow was stopped and then restarted. It is now presenting the following errors in Datasphere and S/4 HANA

SAP Datasphere:

S4 (Transaction code DHCDCMON )

Does anyone have suggestions or experience with optimizing the load process for such a large data set?

Thank you in advance for your guidance!

SAP Datasphere

SAP S/4HANA 

[JulianJuraske]

I would add a Hashtag for S4HANA since the Problem is clearly within the S4HANA System and not the DSP.

[romain_m]

Hello ! Any news about this topic ? 🙂

[bheemeshRao]

Hello @wsamuel ,

Could you please check what is your integrations hours in Datasphere? Configuration -> Tenant Configurations.

Additionally, could maintain the below in your S/4 HANA source and try to execute

[ClemSteich]

Hi,

First thing is to make sure your S/4HANA has all the latest Replication Flow related notes implemented.

Use program CNV_NOTE_ANALYZER_DI to identify them (https://me.sap.com/notes/3016862).

Regards,

Clément

[wsamuel]

Thanks @ClemSteich . We implemented the SAP notes last month but have asked the Basis team to check for any new notes.

[EagleYeSAP]

Hi ,

Please check your CDS views code once and try to optimise the code.

https://help.sap.com/docs/SAP_HANA_PLATFORM/6b94445c94ae495c83a19646e7c3fd56/7b3e645df1d044cead4d208...

 Try to analyze the OOM dump (1984422 - How-To: Analyzing SAP HANA Out-of-memory (OOM) Dumps)

Regards

Srinivas

[wsamuel]

Thank you @EagleYeSAP, i will have a look at both links