Very helpful, thanks for this video! Not sure if I just don't get it, but: On the slide "processing data | conclusion" you're comparing the parquet formatting for Synapse (left) and Fabric (right), correct? On the left / Synapse side you're showing that there's the need for transcoding parquet into vertipaq for Power BI consumption. 1. Is this vertipaq transcoding done at runtime while Power BI queries data in Direct Query Mode? 2. Is this vertipaq transcoding not necessary anymore with fabric? Is parquet data in fabric directly ready for PBI consumption?
@@gucksdu4 No. Not with Direct Query (that one sends queries to a SQL engine). This is with DirectLake. Also, try it out and let us know if you see any performance loss.
Man I love Bogdan!! I will watch ANYTHING he talks about.
How would you compare these enhancements vs. Databricks delta live tables processing or delta tables
So when we say direct lake is that a real direct query from PowerBI or is the data still loaded in PBI dataset?
Very helpful, thanks for this video!
Not sure if I just don't get it, but:
On the slide "processing data | conclusion" you're comparing the parquet formatting for Synapse (left) and Fabric (right), correct? On the left / Synapse side you're showing that there's the need for transcoding parquet into vertipaq for Power BI consumption.
1. Is this vertipaq transcoding done at runtime while Power BI queries data in Direct Query Mode?
2. Is this vertipaq transcoding not necessary anymore with fabric? Is parquet data in fabric directly ready for PBI consumption?
no. Both are Fabric. Left -- this is how we read, via Transcoding, into PBI and SQL. Right -- this is how we write vordered parquet from all engines
Ahh ok thanks!
So with direct query mode, Power BI always needs to transcode at runtime into vertipaq, with potential data processing performance loss?
@@gucksdu4 No. Not with Direct Query (that one sends queries to a SQL engine). This is with DirectLake. Also, try it out and let us know if you see any performance loss.