Register now to learn Fabric in free live sessions led by the best Microsoft experts. From Apr 16 to May 9, in English and Spanish.
We are scratching our heads with this and not sure if we're barking up the wrong tree. And perhaps PBI isn't capable of this. But here goes...
We have multiple customers, each with their own warehouse database. We have an on-prem gateway (installed in AWS) pointing to these various warehouses and have setup the datasources in the PBI service. We would like to publish a standard set of reports to each of these customers, so that they can consume these - ideally via an app.
Our assumptions and observations so far are:
Is there something glaringly obvious we are missing, do we need to consider publishing a template app publicly? Is this something we would need Premium for? Just a lot of unknowns at this point, but it feels a bit like we are trying to use PBI in a way that it perhaps wasn't intended for?
Any advice and guidance would be very welcome!
cheers,
bmenate
We observe a massive impact on dataset refresh performance from poor network architecture, high latency, and network congestion. For us the goal is to reduce network complexity, not increase it with tons of VPNs and traffic bouncing back and forth. YMMV.
fair enough.. we don't have VPNs to think about as the gateway literally sits in the same AZ as the SQL servers.
anyway, thanks for your inputs - am hoping someone can help try answer or shed light on my original post - will check out app controls data stuff you suggested though.
Maybe I am just too hung up on the meaning of the word "on-premise". For me that means the gateway cluster members need to be physically close to the on-premise data sources. In my world this means I place my cluster members in the same physical data center(s) as my main business application databases, for minimum network latency and guaranteed connection uptime.
Of course in your scenario "on-premise" seems to mean "on AWS" so that makes the discussion slightly more academic. I would still love to see a comparison on the network traffic that is caused by such a setup, and its impact on overall performance.
Maybe they should rename "on-premise" to "private cloud".
hey I didn't name the thing 🙂
anyway, isn't it immaterial whether we're in AWS or in your datacenter. You could replace the AWS in my diagram with your DC and the same traversal across the internet to Power BI needs to take place, right?
In my world I place my gateway physically close to the databases - after all, AWS is really also just a bunch of datacenters, not some mythical place where cool stuff happens magically.
There are multiple things that are questionable here. Putting an on-premise gateway on AWS is just ... wrong. It's a travesty.
Anyway. If you have a sufficiently diverse audience for your app then you MUST use separate apps. In your scenario that would mean that each customer MUST get their own workspace and app. You can still feed these workspaces from the same master dataset, but they have to be separate.
Have you considered going the "App owns Data" route of Power BI Embedded instead?
thanks for your reply.
Keen to understand why you think putting an on-premise gateway in AWS is a travesty? We have a classic 3-tier web application with each customer logically seperated by database. The databases are SQL Server running on EC2 - how would you have gone about getting the data to the PBI service out of interest?
I get your advice to ensure each customer has their own workspace and therefore app. However, the problem still remains that we don't have a shared master dataset - each customer is seperate with their own seperate database (warehouse) - but... we want to share the same set of reports to each of them.
I'm not too familiar with the app owns data scenario - but will do some digging. I've attached a basic diagram of how it looks and how we assume it should hang together.
Covering the world! 9:00-10:30 AM Sydney, 4:00-5:30 PM CET (Paris/Berlin), 7:00-8:30 PM Mexico City
Check out the April 2024 Power BI update to learn about new features.