Forum Discussion

pgoligher's avatar
pgoligher
Regular Visitor
4 years ago
Solved

Implications of using two data sources, two gateways, two dataflows and ONE dataset?

Hi all,

 

I have two physical servers, each with their own gateway installed. One is using an ODBC connection and the other is using SQL Server.

 

The problem is: I want to pull data from each server into the same dataset.

 

I have set up two separate dataflows. One dataflow is connected to the ODBC gateway, and the other is connected to the SQL Server gateway. Both dataflows are loading data as expected.

 

With both dataflows in place, I have created a dataset. The dataset is connected to both dataflows and successfully pulling data from both servers.

 

This seems too good to be true. Are there any implications to pulling data from multiple data sources in this way? For example: Performance issues?

 

I want to take everything into consideration before proposing to switch our dataset from a direct SQL connection (as the Progress server is new) to the multiple dataflow/gateway approach.

 

Thanks, Peter.

dataflow
  • ibarrau's avatar
    ibarrau
    4 years ago

    Hi, that is ok and it will work. There is another approach you could use. The dataflows can help building another layer from a transformation purpose that can take the data once and hold it. However they are note completely necessary unless you want for sure to keep that layer.

    The other alternative would be to turn one of the gateways as cluster of gateways, then re install the other one of the gateways in order to make it part of a cluster. Having both of the in a single cluster will make Power Bi believe that it's one gateway that knows the two sources. Then you can start working and connect with Power Bi Desktop (be sure your machine can access both of the sources). After publishing, Power Bi Service will be able to configure the dataset for both sources on a single gateway.

    I hope that helps,

    P/D The best performance for data manipulation will always be in a db engine.

1 Reply

  • ibarrau's avatar
    ibarrau
    Icon for Super User rankSuper User
    4 years ago

    Hi, that is ok and it will work. There is another approach you could use. The dataflows can help building another layer from a transformation purpose that can take the data once and hold it. However they are note completely necessary unless you want for sure to keep that layer.

    The other alternative would be to turn one of the gateways as cluster of gateways, then re install the other one of the gateways in order to make it part of a cluster. Having both of the in a single cluster will make Power Bi believe that it's one gateway that knows the two sources. Then you can start working and connect with Power Bi Desktop (be sure your machine can access both of the sources). After publishing, Power Bi Service will be able to configure the dataset for both sources on a single gateway.

    I hope that helps,

    P/D The best performance for data manipulation will always be in a db engine.