Forum Discussion

PBI_2K26's avatar
PBI_2K26
Icon for Advocate I rankAdvocate I
2 months ago
Solved

Snowflake connexion too slow

Hello community, I'm currently working on a Power BI project and trying to connect to a Snowflake database. However, I'm experiencing very slow connection and data loading times, even when queryin...
Data Protection
Show and Tell
Tutorial Requests
  • v-kathullac's avatar
    v-kathullac
    2 months ago

    Hi PBI_2K26 ,

     

    Thank you for reaching out to Microsoft Fabric Community Forum.

     

    Based on your issue there are few debug steps which you can try and reslove your issue.

    • Network latency between the Power BI client/service and the Snowflake environment.
    • Snowflake connector or driver version (ensure the latest Power BI Desktop version is being used).
    • Authentication method (OAuth, SSO, external browser authentication, etc.), as repeated authentication or token validation can introduce delays.
    • Gateway performance, if Power BI Service is using an On-premises Data Gateway.
    • Proxy, firewall, VPN, or SSL inspection that may be affecting communication between Power BI and Snowflake.
    • Metadata retrieval from Snowflake. Sometimes loading the list of databases, schemas, and tables can be slower than the actual query execution.
    • Check whether the issue occurs across different networks or machines to rule out a client-specific or network-specific problem

    Thanks,

    Chaithanya.

     

  • PBI_2K26's avatar
    PBI_2K26
    2 months ago

    Hello community,

     

    Thank you all.

    The issue was not resolved, but probably as most of you said, It could be a Network (VPN) issue.

    Thank you.

Dev_Dholakia's avatar
Dev_Dholakia
Icon for Resolver IV rankResolver IV
2 months ago

hi PBI_2K26 ,

 

I've hit this before — annoying one to chase down, but usually solvable.
 
If Snowflake runs the queries fast on its side, the bottleneck is between Power BI and Snowflake — not Snowflake itself.


Things to check:

  • Import vs DirectQuery — DirectQuery fires one query per visual, so pages feel slow even on tiny tables.
  • Test a native SQL in the connector's advanced options. If that's fast, your M steps are breaking query folding — right-click the last Power Query step → "View Native Query" to confirm.
  • Desktop → Options → Data Load — disable background data, type detection, and auto-relationships while testing.
  • Driver version — the Snowflake connector is moving from Simba ODBC to ADBC; old drivers cause slow metadata calls.
  • Network — proxy, VPN, TLS inspection, or region distance between Power BI/gateway and Snowflake adds real latency.
  • On the Service, make sure the on-prem gateway (if used) is in the same region as Snowflake. Pure cloud + SSO usually doesn't need one.
  • Check the warehouse isn't suspending/resuming during refresh — bump auto-suspend to 5–10 min.
  • Compare Snowflake Query History time vs Power BI refresh time — the delta is your PBI/network overhead.

Also: auto-added M params (CreateNavigationProperties, ConnectionTimeout, CommandTimeout = null) in the source step are known to slow Desktop badly — remove and retry.


To narrow it down:

  1. Import or DirectQuery?
  2. Slow on Desktop, Service, or both?
  3. On-prem gateway — same region as Snowflake?


    Disclosure: My earlier response above was drafted with the assistance of an AI tool (Microsoft Copilot) and cross-checked against Microsoft Learn documentation. Sharing this note for transparency as per the Community AI Usage Policy