With the announcement July 6, Microsoft has merged the number of gateways used in Power BI. The new On-Premises Gateway not only replaces both of those applications, but also replaces the data gateway used to access on premise data for other applications, such as Azure Machine Learning. This reminds me of Lord of the Rings as There is now One Gateway to bring them all. One Gateway to bind them.
Replacing the Old Power BI Gateways with the New On-Premises Data Gateway
When installing the new gateway, the place to look for the new installer is not on some Microsoft Download page, but from within Power BI itself. The new gateway needs to be installed from Power BI, rather than a generic location. As the gateway is providing data for your account, installing it in this fashion will record the gateway used on this account. To download the gateway, login to the Power BI Service and click on the <<insert icon here>> the download menu and select the Power BI Gateway. This gateway is truly the one gateway to rule them all as it contains the ability to install either the On-Premises Data Gateway, which is the direct replacement for the Enterprise Gateway, or the Power BI Personal gateway. You can install both gateways, but you have to pick one to install first. Both gateways can also run on the same device. Once the software is installed, you can convert your existing gateways to the new gateway, so there is no need to enter the configuration twice.
New Gateway, New Security Model
While the personal gateway works in a similar fashion as before bu using a single user’s credentials, the new gateway uses a different security model than was employed in the past. Instead of using the credentials of the Id which installed the software, it uses a new service account NT SERVICE\PBIEgwService. This account is for the gateway only, it does not need to be granted any data access. The new application is creating a connection to the Azure Service Bus, acts as the clearing hours for all data requests, including Power BI and other applications occur. The application gateway is where the credentials for accessing the data are applied. This security model is used not only for Power BI, but for other applications like Azure Machine Learning where there is a need to get data from a on-premises source and load it to the cloud. Previously it was not possible to have the gateway used for Power BI to be loaded to the same machine as the other data gateway that was used for Azure Machine Learning. Now that they use the same gateway, this restriction is removed.
Functional Differences in the New Gateway and the Previous Gateways
The underlying security model and application is different, but operationally are the new gateway does not behave differently from the Enterprise and Personal Gateways Power BI, there is just a new application do to what they did. If you want to use the data access rights of a single specific user to refresh and update data, the Personal Gateway will do that. The Enterprise Model provides for the ability to inherit the security of the user accessing the data. If you have a tabular model stored locally which is used as a data source for Power BI, the users will see the data they were granted access to from within the tabular model. The scheduled update features are not changed and are still accessed from the same area as they are in Power BI Service. Unfortunately, if you are interested in connecting to Azure Machine Learning or Data Factory, you will still need a different gateway, and that gateway cannot be installed on the same PC as a Power BI Gateway. I will have more information on that gateway in a later post.
Data aficionado et SQL Raconteur