Forum Discussion
Helper ISynapse Realtime analytics
Hi,
If your problem is about sharing the project with other developers, you can use the format PBIP and use a source control repository. The dataset and reports are supported on this format and this would solve your problem.Kind Regards,
Dennes
Helper IHi Dennes,
In terms of architecture:
In our use case, we would like to show the data coming from events streams ASAP. Its a bit of a combination of both, we are receiving events per item, and would like to store the last event per item, historically. I.e., could be a year a go, or a second ago. Storing all data historically wouldn't be a problem though.
Main importance is to display the latest event ASAP, would you recommend us against using KQL then?
In terms of adding a local model -> this is fine, but we haven't found a proper development/collaboration approach on adjusting/changing this local model afterwards, as overwriting existing models in the fabric environment resulted in unknown states where the models/datasets are no longer visualizable.
Which solution fits our described requirements above best?
kind regards,
Kjell
Power ParticipantHi,
I'm not recommending against KQL, in the opposite. KQL is great for real time.
But EventStream is capable to have multiple destinations. At the same time it sends the data to KQL, it sends the data to lakehouse as well. You can use KQL for real time reporting, and use lakehouse for historical reporting without having to cross the data from one to another.
The real time data you discard, it's only for the moment. The lakehouse data you keep for historical purposes according your needs.
In this way, you don't need to have cross access between KQL and Lakehouse. Or at least will minimize it.
Kind Regards,
Dennes