Symptom
Transformation Flow from File space to Storage Space has performance issue.
Environment
- Product: SAP Datasphere
- Transformation Flows
Reproducing the Issue
- Run a Transformation Flow that reads from the object store/file space and writes to a storage space table.
- Apply a filter on a date column and configure batching (for example, very small batch sizes on a date column that create many batches).
- Execute the flow and observe long per-batch runtimes and high total duration, especially when many small batches are generated.
Cause
- The object store table was optimized with Z-order on high-cardinality key columns not aligned with the Transformation Flow’s filter (date), preventing effective file skipping and causing large file scans from the data lake.
- Very small batch sizes led to many batches and high orchestration overhead per batch, increasing total runtime.
Resolution
Please consider the following solutions:
1. Enable the Performance-Optimized setting for the Transformation Flow.
See SAP Help: Change Transformation Flow Settings | SAP Help Portal.
2. Use a Graphical View as the source instead of a SQL View for the Transformation Flow according to KBA
3732192 Transformation Flow loading from SQL View uses excessive memory compared to Graphical View - SAP Datasphere
3. Increase batch size for simple load-focused flows to reduce the number of batches and orchestration overhead; avoid very small batch sizes that produce many batches. Balance throughput against memory/time limits.
4. Align Z-order with the query filter: re-OPTIMIZE the data in the object store with Z-order on the date column used in filters (for example, posting date) and optionally a secondary selective column (for example, company code). Ensure the Transformation Flow filter predicates target the same columns used for Z-order so that file skipping can occur.
See SAP Help: Define Z-Order Columns While Running an Optimize Task.
If performance remains insufficient after applying the above, refer to the documents below and contact support or consulting for further tuning guidance.
See Also
- Refer to: Note 3732192 - Transformation Flow loading from SQL View uses excessive memory compared to Graphical View
- Refer to: Note 3538038 - Important considerations for SAP Datasphere object store
- Refer to: Note 2626344 - Who is the SAP Account Manager/Executive for my company?
- Refer to: Change Transformation Flow Settings
- Refer to: Optimize with Z-order in SAP Datasphere
Keywords
Datasphere, Transformation Flow, object store, file space, storage space, performance, slow load, batch size, packaging, Z-order, optimize, posting date filter, data lake, file scan, Graphical View, Performance-Optimized , KBA , DS-DI-TRF , Transformation Flows , How To
SAP Knowledge Base Article - Public