Does dataset join location affect SPICE capacity usage?

Assuming one needs to join a large measure table with a dimension table to get additional fields for each measure record (assuming a many-to-1 join).
With reference to the 1TB SPICE limit per dataset, is there any benefit in ingesting both tables separately into SPICE and then doing the join on QuickSight versus joining both tables in one query and ingesting it SPICE? - would it use less SPICE storage if the join is done on QuickSight rather than on the data source or does it not make any difference?


1 Like


In SPICE you can bring dataset from different heterogeneous sources and join it. If all tables participating in join belong to same database (datasource) you may join prior. Database level join help in removing unneeded columns prior in SQL query projection and may save space in SPICE dataset once loaded.


Thanks @AnwarAli . It is a very good point, if the data sources are different, the SPICE is a good approach , by saying that it will be good to select desired columns ( which is require for the analysis and joining condition) and then do the join at SPICE level. If the data source is same, best way to have a custom sql and select the right column set from both tables and do the join and then take to SPICE.

Regards - Sanjeeb