I have 120 columns in CSV format but each visual uses only some of those, not all columns. With calculated fields existing on some columns only, I see visual loads slow. Some calculated fields take 1 min to load. The dataSet is in SPICE. The data size is about 10GB and # of rows is around 3-4M.
How to make visuals run faster in this case? Can converting it to Parquet format help here?
Hi @tbdori - are you creating visuals as table type? Since the data set is already in SPICE, QuickSight should be faster. 3-4M rows is very high for a reporting. We should build some graphs, aggregation on top of the data.
Please share some screenshots and details what exactly you are trying to do so that it will help you in guiding the right solution.
Tagging QuickSight experts for their feedback.
@Max @Thomas @thomask @Bhasi_Mehta @apjvinod @Kristin @Tatyana_Yakushev @David_Wong @Biswajit_1993
Regards - San
Hi @tbdori, Need to follow some rule to fast your analysis.
Replace all you calculated field from the analysis level to dataset level
Restrict the use of minimal Table Visual & Pivot visual because in this case you are adding so many field so QuickSight take time to loading these field.
Restrict the use of Distinct Count() for String type data.
Restrict the use of sheet level filter.(Instead of this you can use the calculated field)
Use Parameter Control filter instead of Column level filter.
These are some basic things to maintain in our day to day dashboard building time.
PFB some tips in details how to improve dashboard performance.
Thanks & Regards
Biswajit Dash
Thanks, here are more accurate details.
CSV file contains 8 columns and 40M rows
So, # of column is not the cause, I guess.
We do have charts and KPIs. Those run OK.
But, slowness is shown on tables that contain several calculated fields like distinct count|percentage of the records based on date value like QuarterOverQuarter.
@Max @Thomas @thomask @Bhasi_Mehta @apjvinod @Kristin @Tatyana_Yakushev @David_Wong @Biswajit_1993
