FYI: @Giorgos Malandrakis Hi, we enabled the mcp for a couple of internal users and it has been really good so far. But, we can see that it sometimes generates really massive SQL queries to Snowflake. Is it possible to have a specific Gooddata user to use its own warehouse (compute cluster) in Snowflake? We want to isolate the queries from the mcp to its own warehouse basically.

Great to hear the MCP integration is working well for your team! Regarding your question about isolating MCP-generated queries to a dedicated Snowflake warehouse — there isn't a built-in GoodData feature to route queries based on a specific user to a specific Snowflake warehouse. However, there are two relevant approaches worth considering: 1. Data Source Routing (Workload Isolation by Dataset) GoodData's Data Source Routing feature lets you route analytics queries to different compute clusters (i.e., different Snowflake warehouses) based on the datasets involved in a query — not based on the user. You can set up: • A primary data source (e.g., a smaller warehouse for pre-aggregated/dashboard workloads) • An alternative data source (e.g., a larger warehouse for heavy, detailed queries) GoodData automatically routes queries to the appropriate cluster depending on which datasets are needed. [Data Source Routing] This could help isolate heavy MCP-generated queries if they consistently touch specific datasets that you can assign to a dedicated warehouse. However, it's dataset-driven, not user-driven. 2. Snowflake role Parameter (Per Data Source) When creating a Snowflake data source in GoodData, you can specify a Snowflake role and warehouse via the

parameters

field in the data source API. For example:

"parameters": {
  "role": "<SNOWFLAKE_ROLE>",
  "warehouse": "<SNOWFLAKE_WAREHOUSE>"
}

[Permitted Parameters] If you created a separate GoodData data source pointing to a dedicated Snowflake warehouse, and configured your MCP users' workspace to use that data source, you could effectively isolate their queries. However, this would mean managing a separate data source configuration, and the routing would still be at the workspace/data source level — not per individual user. Summary The knowledge sources do not describe a native mechanism to route queries to a specific Snowflake warehouse based on the GoodData user making the request. If this is a critical requirement, I'd recommend reaching out to GoodData support to discuss whether a workspace-level separation or a roadmap feature could address your use case. 👍 0 button 👎 0 button ℹ️ Tip: Mention @kapa.ai in the thread for follow-ups. 👀 Our Support team monitors all messages and will jump in if needed., with interactive elements

Hi Philip, Regarding isolating MCP-generated queries to a dedicated Snowflake warehouse, there are a couple of options to consider. One option would be data source routing as was mentioned above. However, this is not configurable on a per-user basis, and the routing decision is handled automatically by the system. A more suitable approach would likely be to create a separate workspace with a separate data source configured for it. That data source could use a dedicated Snowflake warehouse intended specifically for the MCP technical user or MCP-related workloads. This would allow you to isolate those queries from the warehouse used by your regular users. So, while assigning a specific Snowflake warehouse directly per GoodData user is not currently supported, separating the MCP use case into its own workspace and data source should provide the isolation as a workaround.