08-17-2018 05:05 AM
This is because of you connection is established directly to your dataset. Microsoft recomends to build tabular model, so that your data will be stored in olap engine and will throw results alone to your PBI Reports.
08-17-2018 06:53 AM
I'm not sure I fully understand. The way I'm pulling data (and I bet the others in the tread are) is to hook up via ODBC to hive and run an import on a table or a query that results in a "tablular" dataset. As far as Power BI should be concerned, it's the same output as hooking up to an oracle table using ODBC, or any other database.
Is the tabular model something different then an output that looks like a table (column/rows)?
10-05-2018 12:28 PM
This is simply unacceptable. This is the definition of death by a thousand cuts. We have over 60 dashboards spread across 3 servers and that's still not enough to off load the issue this is creating. If a dashboard is pulling in 5 tables thats 10 connections and executions to pull one dashboard.
In the examply below I ran a trace on the dashboard refresh and you can clearly see the query running twice with a connection reset inbetween. It's impossible to negate the negative effect of this many connections even to multiple servers.
10-15-2018 06:18 AM
The Problem occurs when Power BI issues a Hive Query something like 'SELECT * FROM DATE' but if imported as the table it uses only one connection. Hope this helps
02-01-2019 07:52 AM
The repeated running of the same query is a pain that affects multiple sources.
I raised an idea:
If you feel strongly, vote for it.