01-30-2024 11:17 AM
Hey all,
We are currently looking into BI Blend and use it for our Sensible Machine Learning solution that we are about to implement.
The initial plan is to use two BI Blend instances one for historical data which is very detailed data with a long historical range.
And a second for a more short term forecast period.
Our ambition long term is to be able to use our Forecast BiBlend table data in our Planning process and our regular Cube and Scenario.
Do anyone have a recommended approach to migrate data from BIBlend into cubes?
Solved! Go to Solution.
01-31-2024 09:25 AM
Using a connector for a data source is the same as any other data source used in the workflow (on a very simplified level), so yes, you would typically use this in an import step in the workflow. You could also consider a direct load instead which would be faster while taking away some of the auditing capabilities (amongst other things). But a direct load is also using an import step in the workflow.
You will need to use workflow specific parameters to ensure that the connector pulls the data from the correct table. But that is part of the normal configuration when setting up a connector (rule).
No, unfortunately I do not have an example that I can share here for this exact purpose. I would need to create one myself.
01-31-2024 03:42 AM - edited 01-31-2024 03:47 AM
Hi,
Those are very deep questions that cannot be properly answered here in this forum and should typically be part of your SML project's detailed design discussion and documentation.
A few high-level comments.
I am not sure about the suggested data flow here. Do you mean to use one BI Blend table for you historical data which then flows into SML, which then flows into the other BI Blend table that holds the forecast data, which should then be copied from BI Blend to the cube's forecasting scenario? Though BI Blend can be used as a source table for SML, it does not necessarily need to be used. During the design, this question will typically be addressed and then all stakeholders will decide whether to use BI Blend or not.
Also, BI Blend is driven by the workflow as well, so I assume that even the "long historical range" will be loaded to different periods. This has an effect on how you can report the data as it is table based and not cube-based.
As for your actual question, there are - as always - different possibilities. You could e.g. set up a data source that loads the data from BI Blend into the cube via an import step in the workflow or use a business rule that is executed from a Data Management step or a dashboard button. Generally I prefer the import step solution as it is easier to maintain, more transparent and easier to understand for the end user. However, before 'just' using this answer to load data from BI Blend to the cube, make sure you scrutinize the intended design with the appropriate time and resources to ensure the process is well designed as well.
01-31-2024 04:22 AM
Hey Henning,
Thanks so much for your input!
I agree they are deep and at the moment mainly from an solution evaluation perspective on possible future needs. Since It's my first time working with BIBlend I try to see if there are smart ways handling the data. I did read through the DesignReference guide but did not fins any direct answers to this question in particular.
Could you elaborate a little bit on the data source solution that you mentioned. We have done similar before with Data Mgmt Export Sequence and I agree with you that they are preferred from a maintain perspective. However It is not clear for me how this can be used for BIBlend, or do I miss somthing? When trying to do a regular DataManagement Export I get no data since the BiBlend data is table based and not cube based as you say.
Thanks
Jones
01-31-2024 05:41 AM
Hi Jones,
BI Blend stores data in a table (as you correctly pointed out) that you can connect to just like using other configured internal or external source tables.
To connect to this you will need to create a Connector Business Rule and then use the Connectors Data Source to connect to your table and load the data via your workflow.
In your connector data source you can then select your connector BR:
01-31-2024 09:17 AM
Thanks for the images.
Will the data export through connector be executed from the Workflow? is it supposed to be a regular Import, Validate, Load? And should any setting apply to workflow in order to fetch the data from the workflow?
Also do you have any business rule examples that could be used to fetch the specific bi-blend table?
Thanks
Jones
01-31-2024 07:20 AM - edited 01-31-2024 09:17 AM
01-31-2024 09:25 AM
Using a connector for a data source is the same as any other data source used in the workflow (on a very simplified level), so yes, you would typically use this in an import step in the workflow. You could also consider a direct load instead which would be faster while taking away some of the auditing capabilities (amongst other things). But a direct load is also using an import step in the workflow.
You will need to use workflow specific parameters to ensure that the connector pulls the data from the correct table. But that is part of the normal configuration when setting up a connector (rule).
No, unfortunately I do not have an example that I can share here for this exact purpose. I would need to create one myself.
02-01-2024 03:49 AM
Many thanks Henning. Your insights are very helpful having with us if we would have to go this route in the future. Cheers