We are establishing a new blog series around SAP Integrated Business Planning (IBP) and Integration via SAP Cloud Platform Integration – Data Services (CPI-DS). You can imagine it is like the “tip of the month” where we will publish a series of posts on how to best integrate data from and to SAP IBP.
Today we will start with the Extraction of Data from IBP via CPI-DS. Did you ever struggle with Data Integration Jobs that randomly fail? Error message: Error: <“curl_easy_perform”:52:Server returned nothing (no headers, no data)>.
What is the reason?
If you are extracting Key Figure data from IBP to another System, the data is aggregated, calculated and processed by the IBP calculation scenarios. The calculation scenarios can be big depending on the amount of data is read and complexity of the key figure calculations (L-Code). Due to the architecture of the system the data extraction jobs have a timeout of 10 minutes. If a calculation scenario query doesn’t finish within 10 minutes, the connection breaks and no data is passed. The curl error appears.
Occasional problems: Someday data can be extracted within that limit, some other days not. What is the reason for it? The runtime of the query does depend on overall system load by other IBP activities.
What can you do?
Tip 1: Instead of running 1 big job to extract data, use multiple jobs.
Split the size of the job by applying filters and schedule jobs in a sequence.
E.g. Filter for certain Locations or Products/ Families / Business Segments.
Furthermore filter on time horizon and extract at needed weekly/monthly/… granularity.
It is important the filter applied in CPI is passed to IBP calculation scenario: How can you check that?
On the CPI Data Flow: goto View History -> Monitor Log
compare the row count between top line and bottom line. (note: there can be multiple lines) If there is a difference you should have a closer look on the filters and transformation steps in the CPI Data Flow.
Tip 2: reduce the number of Key Figures
if the data extraction job is running long, you can also split by Key figures. One troubleshooting guide is to separate stored Key Figures from those which have a complex calculation logic.
- Be careful, when extracting data of key figures
- that have an L-Code calculation on
- external Key Figures (part of order-based planning)
Tip 3: plan early for performance testing
Make sure that you do have proper performance testing of the extraction prior productive use / GoLive of the system. Consider that over the time the planning area size may increase and that the extraction still is well below the 10 min timeout.
Important SAP notes:
|2493042||IBP- HCI Integration – Best Practices for extracting data out of IBP|
|2685841||CPI-DS task error: Error: <“curl_easy_perform”:52:Server returned nothing (no headers, no data)>|
- In a follow-up blog we will detail out how to best filter data?
I’m interested on your feedback, please let me know.