3796100 - How pagesize controls the extraction window in Scripted (Advanced) Extraction using Change Number Range partitioning in SAP Signavio Process Intelligence

• When using Scripted (Advanced) Extraction with a Change Number Range partition in SAP Signavio Process Intelligence, it is unclear what the pagesize parameter controls and how it affects the delta cursor, row counts, and Remote Function Call (RFC) volume.
• Delta runs return far fewer rows than the initial load and it is unclear whether this is expected behaviour.
• After a period with no change activity, extraction stops even though records with higher change numbers exist further along in the table.
• It is unclear whether changing the pagesize value between runs will cause data loss or gaps in the extracted data.

1. What pagesize controls

In Change Number Range partitioning (type: id), pagesize defines the width of the numeric range queried per run — it is not a row count limit. Each run queries the source table using the condition:

WHERE changenr >= <initial> AND changenr < <initial> + <pagesize>

The number of rows returned depends on how many records fall within that window, not on the pagesize value itself.

2. How the delta cursor advances

After each run, the cursor is saved as initial + pagesize (the end of the queried window), regardless of how many rows were returned. On the next run, initial is set to this saved value and the window advances by another pagesize.

• If a window returns zero rows, the cursor still advances to initial + pagesize.
• The next run begins at the end of the empty window — no range is re-queried.

3. Row count on delta runs vs initial load

Delta runs are expected to return significantly fewer rows than the initial load. Once the historical data has been loaded, each delta run only covers the narrow window between the current cursor position and current cursor + pagesize. At the leading edge of the data, this window spans only recent incremental changes.

4. Empty-window stop behaviour

If the pagesize window is narrow and falls in a period with no change activity, extraction stops at that point. It does not skip ahead to the next non-empty window. To span quiet periods reliably, use a pagesize wide enough to cover typical gaps in change number activity.

5. Remote Function Call volume trade-off

A wider pagesize means more RFC calls per run, because the SAP ERP JDBC driver paginates each window using a fixed page size (default: 50,000 rows per RFC call). For tables with many partition combinations, a very wide pagesize can generate a large number of concurrent connections. If this approaches gateway conversation limits, reducing pagesize will lower the RFC call volume per run — though this does not resolve an underlying gateway capacity constraint.

6. Changing pagesize between runs

Changing the pagesize value between runs does not cause data loss or gaps. Because the cursor is always saved at initial + pagesize (window-end), the new run begins exactly where the previous window ended. No range is skipped and no already-loaded data is affected.

https://help.sap.com/docs/signavio-process-intelligence/onboarding-and-data-integration-guide/partitioning-based-on-change-number-range?locale=en-US&version=LATEST

soe, scripted extraction, advanced extraction, pagesize, page size, CHANGENR, change number, change document, delta, CDPOS, CDHDR, partition, range, empty window, RFC call volume, PINT, Process Intelligence , KBA , BPI-SIG-PI-DS , Data Source (CSV, Connector) , How To