Understanding Gateway Clusters in Microsoft Fabric

[@tlanza89]

I would love a video on how to set which gateway should be the primary in a cluster!

[@daryllynch998]

Thanks Adam. It would be nice to see video of the VNet Gateway set up with advice on Capacity size and region. e.g. do we need to consider a cluster for VNet Data Gateway?

[@Mhmod48]

Thanks Adam Is there a rule of thumb about how many on-premise data-gateways one needs in a cluster to connect to a certain amount of on-prem data?

[@trevorkincaid5059]

Is there a way to migrate a personal gateway to an on-premises gateway? Could you make a video on that?

[@agnego]

Hi! Q- We have an on-prem data gateway cluster, set up long time ago. The recovery key is rather insecure and we would like to change it but we have many existing datasources and MS advice is that credentials for all datasources would have to be re-entered if the recovery key is changed. We did some testing with web api data source and it didn’t seem to be the case - we didn’t need to re-enter credentials, however, we cannot test any other sources such as SQL, AD, etc. Have you ever come across this? Would appreciate your advice. Thanks!

[@bertholdflikkema8599]

Q: the gateway cluster can also be used in fabric datafactory right? What I'm curious about is how the concurrency settings of the gateway in fabric datafactory (assuming those exist, same as in azure datafactory) relate to concurrency settings inside pipelines. For example, let's say your gateway is configured to handle 5 concurrent jobs, one pipeline that uses the gateway has 10 concurrent activities, does that mean 5 activities will have to wait? (Assuming the server the gateway is installed on is having an easy time and is not the bottleneck) In other words; does the number of concurrent jobs relate to the number of concurrent pipeline runs or to the number of concurrent activity runs? MS documentation for the shelf hosted integration runtime is not clear on this. And also, how should the number of concurrent jobs be set in relation to the processors/threads of the machine(s) it is installed on.

[@RichardOliver]

Our nodes in a cluster are reporting different version numbers, bizarrely, despite both being updated each time (currently 3000.222.5 and 3000.186.18).

[@davidcousins8407]

Yep, been like that for a while. If you login to the gateway it sometimes triggers the service to realise that you have indeed updated it.

[@franciscorodriguez255]

Excelent !