I'm having trouble working with multiple data sources While our datasource is being developed, I've been using the provided GD Snowflake datasource. This has worked well for some sample data as well as working through some of the embedding implementation However, I've noticed that in the

Data

tab, where you an create Logical Data Models, I'm getting some errors when switching between data sources. If I have the GD Snowflake datasource selected, the LDM seems find (2nd Image). If I switch the selected data source our our own datasource. Those LDMs show errors (3rd Image). If I have the non GD datasource selected, the insights error as you might expect (4th image). This is confusing to me because the GD University videos explicitly call out multiple data sources (5th Image) https://university.gooddata.com/data-loading-overview/753781 I'm hoping I'm missing something. Can someone confirm that we can have multiple datasources for a workspace and explain what I've done wrong?

Hi Levi, I can confirm that GoodData Cloud does not currently support using multiple datasources within single LDM in the Workspace. Thus the errors you are seeing. In any case, you are not the first one asking about this functionality. Therefore, I will submit a product feedback on your behalf so our product team can consider adding this into our future developments. Thank you very much for sharing your experience with us!

🎉 New note created.

Question: does your Aurora contain the same physical model which is in "our" Snowflake? (order_lines, campaigns, ...)

It does not contain the same PDM as the Snowflake PDM. The Aurora datasource is the one that we are building as a data warehouse. I'm using the Snowflake datasource while we finish building the our data warehouse.

When you switch to different data source in LDM Modeler app, all mappings from LDM entities (datasets, attributes, facts, ...) are reloaded. If the target data source does not contain the same PDM entities, the mapping is cleared and then if you execute a report, if fails, because we cannot map LDM entities to PDM entities and generate SQL. If the target data source is completely different, clear the LDM and build it again. If it is sort of similar, you can just edit a few "broken" mappings. When you switch to Edit mode, there is a collapsed panel on the right side, where all errors (including the missing mappings) are displayed.

Fair enough. This checks out with my experience. I was under the impression that I could have multiple datasources and when I drug those tables onto the canvas, they would maintain the reference to the datasource. I thought the datasource selector was a convenience for viewing, not impacting underlying functionality.

Well, this is not yet supported as Julius mentioned. We are thinking about to provide something like this in the future. Now you have to use a data source federation technology, we support Apache Drill and Dremio.

Luckily it is not required by our current use case. We will only have a single data warehouse to pull from. Its only been confusing while trying to build the insights off of the Snowflake data and build the Aurora datasource LDMs in the same workspace.