I don't know if you're still looking for a solution to this @JonasN but I found something that works well for us. We are doing this to get all of our application insights data into an Eventhouse pretty cleanly.
1. Setup an Event Hub instance if you don't have one already in Azure.
2. Go into your log analytics instance and go to Data Export. Setup a data expor to that event hub. You can specify which table(s) you want to have exported. Set the event hub as the destination.
Now you have the data going to the event hub and eventhouses are made to easily ingest that data.
Go into Eventhouse(s) in Fabric, create a new kql db if you need a new one, and go to "Get Data". PIck that event hub. Here is where things get tricky.
KQL is amazing at data ingestion - but eventhub data will come in as an array; and you cannot define an ingestion mapping for a KQL table that will explode arrays, so you ahve to do it in 2 steps.
1. Create a really really basic table with one column named records of type dynamic.
2. Set that table as the destination for the ingestion. Let's call it LogAnalyticsIngestion_Raw
3. Define the tables you want to store the data into. You can expedite this by using Data explorer to connect to the log analytics db, and copy the schemas from there (presuming you want them to be the same)
4. Setup an update policy on each table that will query records, run mv-expand to explode the array, and pull out the data you need for X table. Do that for each table.
This worked well for us, and it allowed us to do things like - -- take any synthentic AppPageViews and dump them into a separate table with a shorter retention period,.... ditch all SQL AppDependencies altogether, and so on.
