Will migrating from Google Spreadsheet to other DataBases speed up the App?

[Ricky_Chua]

We have been using AppSheet linked to Google Spreadsheet as a database. There are significant volumes of data currently stored on the spreadsheet, going well into 15,000 rows in one of the tables. AppSheet is really slowing down for us, and this is after removing nearly all reference virtual columns to speed up (which showed a noticeable improvement).

Has anyone here used another "DataBase" other than Google Spreadsheet? Apologies if i'm not using the correct terminology as i'm self-taught. Will AppSheet speed up if we migrate our data to, say, a Cloud Server like SQLServer or Postgre? (I have never used those before, i'm exploring options)

I know that AppSheet DataBase is about to launch soon, but there have been no updates on speed. We can't test the preview version properly for speed comparisons either because we have >10,000 rows. 

Thank you for any advise or suggestions!

[Rafael_ANEIC-PY]

Hello there, reference columns are not the illness that plagues your app, but the cure, they are the most effective method available for making any kind of calculations, what poisons your app are (probably) virtual columns with expensive expressions (selects, lookups, etc).

And if that the case, I don't think switching the database would help, as the virtual columns I have mentioned are calculated by AppSheet server anyways.

You can have 10k rows if you want, but with how many columns? and in that table, are there any virtual columns? or in other tables, are there any virtual columns that run dynamic calculations based on the large table?

[Ricky_Chua]

Thanks for the reply. It was a typo on my part. I meant to say we have removed nearly all virtual columns, especially those with SELECT and LOOKUP functions. 

There are some tables with over 30 columns, and unfortunately one of those tables is the one with over 10k rows. But that table has only 5 virtual columns, all of which are referencing a single value from another table (eg: [ABC Table].[XYZ Column]).

[AleksiAlkio]

Virtual ref columns have an affect, but not so dramatically as Rafael said earlier. BUT.. they have. So.. if your app will be a little bit bigger, it's better to use Enum/EnumList base type of Ref from the beginning.

For the SQL, you would need Appsheet's Enterprice subscription which change the number of parallel. Then the app reads 10 tables at the same time, in theory. Though it depends do you have relations between your tables. But the biggest advance in performance is coming from the Security filters if you are able to write them in a simple way so the filtering will happen directly in the DB. That means it only sends those rows to AppSheet server. When using GSheet as a data source, it sends all rows from all tables first to AppSheet server, and then security filter filters rows away that are not meant to send to device.

When the row amount is filtered, you have less virtual calculations needed as well.