Performance differences between lakehouse SQL endpoint DirectQuery vs ADLG2 lakehouse connection

[Element115]

Assume one could directly connect to the lakehouse table using the Azure Data Lake Gen2 connector in DirectQuery mode, just like one can presently connect to the same lakehouse table through the SQL analytics endpoint of the lakehouse.

Question:  Will the end user of the report see any difference in data load time when using the report slicers linked to this table?  In other words, is one connection more performant that the other, or is it basically the same?  

I can't test it because the first way using ADLS2 fails with an error message.  But in case someone, or Microsoft has been able to test it, I would like to know what the deal is please.

[amitchandak]

@Element115 , When you store the data in Lakehouse in a structured format, it is saved in delta parquet format, optimized for data retrieval.  As the underlying store is ADLS Gen2 both for lakehouse and ADLS Gen2. The tools will give the same performance for the same type of data. But Fabric engines are optimized to work on this environment

Shortcuts are possible from ALDS Gen2 to Fabric

Azure to Fabric - https://youtu.be/KIbt9GzAdWk

[Element115]

Thanks for the reply, but could you please refine your answer?  Why are you mentioning 'Fabric engines'?  What is an 'engine' in this context? Do you mean there is a SQL Server engine behind the SQL analytics endpoint?

And what do you mean by: 'As the underlying store is ADLS Gen2 both for lakehouse and ADLS Gen2'?  Why is ADLS Gen2 repeated twice? It's not clear what you mean.