SAP Knowledge Base Article - Public

3658925 - SAP Analytics Cloud Data Action Performance: Troubleshooting and Best Practices

Symptom

Common contributing factors are: 

  • Large data volumes being processed
  • Complex advanced formulas or multi-step logic
  • Inefficient model design
  • Broad filter scopes
  • Insufficient system resource

Environment

SAP Analytics Cloud

Cause

Most common causes include:

  • 3128225 - Memory issues when doing planning operations in SAP Analytics Cloud
  • 2811544 - Error "An unexpected technical error occurred and has been logged." after executing data action" occurs after executing data action in SAP Analytics Cloud

Resolution

  1. Identify the Symptoms 
    • Is the data action taking unusually long for specific versions or dimensions?
    • Is the issue always present or intermittent?
    • When did the issue start occurring?
    • Is this happening when triggered from a story or during scheduling?

  2. Use the Performance Analysis tool to identify where time is being spent. See SAP Analytics Cloud Data Action Performance Statistics and Analysis located at System/Common/SAC Content/SAC Content Performance/SAC Content Performance - Stories

  3. Review Data Action Monitor. 
    • Which step of Data Action is taking long?
    • Errors or warnings
    • The size of the data in the step which takes a long time to complete

  4. Review Each Step of the Data Action
    • Copy Steps
      • Check the source and target scope – are they too broad?
      • Filter and narrow the scope as much as possible
      • Avoid copying large volumes of unnecessary data
      • Avoid using calculated accounts or accounts with exception aggregation as source members if possible. See Copy Steps with Calculations and Exception Aggregation.
    • Advanced Formulas
      • Can logic be simplified? See Optimize Advanced Formulas for Better Performance & Managing Data Action Performance - SAP Learning
        • Reduce the Number of FOREACH functions
          • For example, using FOREACH.BOOKED [d/COSTCENTER], there is 10000 iteration. Using FOREACH.BOOKED [d/COSTCENTER].[p/PROFITCENTER], there is 500 iterations. Then consider using FOREACH.BOOKED [d/COSTCENTER].[p/PROFITCENTER] .
        • Remove unnecessary IF statements and organize IF statements to reduce the calculation scope
        • Reduce the number of RESULTLOOKUP functions used in a single calculation
        • Avoid using the "= NULL" comparison if possible
        • Avoid using the "!=" (not equal operetor) for large member dimension
          • For example, replace MEMBERSET [d/COSTCENTER] != "#" with IF [d/Z_CO_COSTCENTER] != "#" THEN .
        • Reduce the number of DATA functions
        • Use dimension aggregation functions where possible
        • Define as many MEMBERSET statements as possible
        • Try splitting complex logic into multiple data actions. 

  5. Try scheduling your Data Actions or Multi Actions during off-peak hours when fewer users are active in your tenant. This way, more system resources are available to power through the task smoothly and efficiently.

  6. Review the version size - Is the slow performance due to the large public edit version size?
  7. Review the Model Design - Is the model set up efficiently?
  8. Testing Data Action in Pre-Production - It is essential to test Data Actions in a Pre-Production environment to validate their performance and logic.
    Please note that you might have a smaller system for Pre-Production than Production. The same performance and behaviour should not be expected in the test environment as in production if the systems are not equivalent(for example, you are testing your data action with the production level data on a 128GB test system, and your production is 512GB).
    • Ensure that the data used is appropriate for and tailored to the system in use.
    • Maintain identical model structures between Pre-Prod and Prod.
    • Utilize the Data Action Monitor to review execution time, scope, and results.

Please note Public Tenants section of Lifecycle Management in SAP Analytics Cloud in SAP Analytics Cloud Help.
"Public tenants share an SAP HANA system with other customers, so there’s no way to directly control the amount of memory or storage dedicated to one tenant. If you need explicit control over the amount of memory or storage, you should license a dedicated private test tenant instead. This tenant can be used for productive or non-productive use."

See Also

Your feedback is important to help us improve our knowledge base.

Keywords

Memory, unexpected technical error, preparing to execute, advanced formula, optimize, data action, performance, slow. performance statistics, analysis, memory issue, planning operation, FOREACH, RESULTLOOKUP, DATA, MEMBERSET, Audit, Data Lock, data volume, insufficient resource , KBA , LOD-ANA-PL-DA , Data Actions , LOD-ANA-PL-AF , Advanced Formulas , LOD-ANA-PL-MA , Multi Actions , LOD-ANA-PL , Planning , LOD-ANA-PL-JM , Planning Job Monitor , LOD-ANA-PL-VER , Version Management , How To

Product

SAP Analytics Cloud 1.0