Forum Discussion
Microsoft EmployeeInitial large data sets in semantic models is resulting in Resource Governing: Memory error
Hi Sreeja_G ,
As lbendlin said, there seems to be an ambiguity in the problem you are having. But I have given some suggestions in my own way of understanding:
Issue 1: Resource Governing: Query has execeeded the memory limit
Each capacity has an effective memory limit based on its SKU (e.g., P1, P2, P3), which specifies the maximum amount of memory that can be consumed by a single operation. This limit ensures fair resource allocation and prevents any single operation from monopolising capacity resources.The reason you can load smaller datasets without hitting this limit is likely because smaller queries consume less memory, staying within the limits of available resources. When you try to load a large table all at once, the memory required for the operation exceeds the effective memory limit, triggering an error.
To mitigate this problem, consider the following steps:
- Optimise the data model to reduce the memory footprint. This may involve filtering unnecessary columns or rows, optimising data types or aggregating data at a higher level.
- Do not load large tables all at once. Use incremental loading to break the data load into smaller, manageable chunks.
- If you experience these limitations frequently, consider upgrading to a premium capacity SKU to increase the effective memory limit.
Issue 2: userprincipalname()
If you find that running a DAX query defaults to a previously tested alias, even if it is not explicitly mentioned, this may have something to do with how the Power BI service caches the query or user context to optimise performance. It is also possible that the test environment still retains some user context that affects subsequent queries.To further diagnose this behaviour, consider:
- Clearing any caches in your Power BI environment.
- Ensure that your DAX queries explicitly define user contexts where necessary to avoid ambiguity.
Best Regards,
Ada Wang
If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.
Hi Sreeja_G ,
As lbendlin said, there seems to be an ambiguity in the problem you are having. But I have given some suggestions in my own way of understanding:
Issue 1: Resource Governing: Query has execeeded the memory limit
Each capacity has an effective memory limit based on its SKU (e.g., P1, P2, P3), which specifies the maximum amount of memory that can be consumed by a single operation. This limit ensures fair resource allocation and prevents any single operation from monopolising capacity resources.
The reason you can load smaller datasets without hitting this limit is likely because smaller queries consume less memory, staying within the limits of available resources. When you try to load a large table all at once, the memory required for the operation exceeds the effective memory limit, triggering an error.
To mitigate this problem, consider the following steps:
- Optimise the data model to reduce the memory footprint. This may involve filtering unnecessary columns or rows, optimising data types or aggregating data at a higher level.
- Do not load large tables all at once. Use incremental loading to break the data load into smaller, manageable chunks.
- If you experience these limitations frequently, consider upgrading to a premium capacity SKU to increase the effective memory limit.
Issue 2: userprincipalname()
If you find that running a DAX query defaults to a previously tested alias, even if it is not explicitly mentioned, this may have something to do with how the Power BI service caches the query or user context to optimise performance. It is also possible that the test environment still retains some user context that affects subsequent queries.
To further diagnose this behaviour, consider:
- Clearing any caches in your Power BI environment.
- Ensure that your DAX queries explicitly define user contexts where necessary to avoid ambiguity.
Best Regards,
Ada Wang
If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.