In this blog we are going to cover 2nd Scenario Data Extraction from CDS views to Native HANA system via SDI
In the first blog we had covered – Data Extraction from BW Extractors to Native HANA system
Requirement – To Extract CDS Views Data from S/4HANA to Native HANA system.
Pre-requisite & ABAP Adapter settings – Already discussed in Part I of the blog.
Detail Steps –
- Create CDS view as shown below – It should have all the annotations needed for Extraction, Delta Fetch and annotation to detect deleted records.
- Check CDS view under Remote Source – You will find it with SQL View Name, in this case ZCDSSOHDRITM$F (Note Dollar($) F stands for Data Category as Fact)
- Create Flow Graph as shown below –
- Flow Graph Source n Target – As discussed in Part I of the blog Source of the flow graph is Virtual Table created on top of CDS view and Target is Column Table.
- Settings of Data Source in Flow Graph –
- Target Table setting in Flow Graph – Upsert mode
- Execute the Flow Graph – Once the Job is executed successfully observe the Subscriptions in ODQMON. I ran the job with delta setting but there are 2 extraction requests – one with init without data transfer and other with full load.
- Check the number of records in source n target – It’s 1782
- Delta test with an example – Transaction VA02
- Sales Order Number – 1716
- Item 10, Material – 8900103
- Quantity – 25
- Check the same in Native HANA table
- Change the Sales Order – New Item 20 is added with Order Qty 45
- Execute the Flow Graph and check the entries in HANA Table
- Change the Order Qty of item 20 to 110
- Check the Delta Entry in ODQMON – 2 records since Order & Item forms are the key in CDS view
- Check the updated value in HANA Table
- Delete the Sales Order Item 20 for the order 1716
- Execute the Flow Graph and check the Delta Queue entry – Item 20 comes is transferred with Order Qty 0.
- Check the Order again in HANA Table
Conclusion – Extraction From CDS views in Native HANA works fine, except that table entries may grow with deleted records. These records can be identified easily and filtered out in the modeling views.
Mayank Jaiswal & Hemant Patidar