Dashboard performance
Should I be expecting the dashboard table view to be functional for large tables or is it more for getting off the ground with smaller datasets
4 Replies
I've been using it to troubleshoot a migration result in a table of 12k rows and ~30 fields, and scrolling, filtering, etc take a really long time. I understand full table scans can hurt the filtering, but 12k rows isn't terribly much. I'd at least hope the UI could avoid locking down when trying to work with filters and such.
I'm guessing this is less about Convex limitations and more about the dashboard not being top priority for the team at the moment, which is super fair (and I would agree with that prioritization fwiw). But wanted to confirm I'm not holding it wrong.
Yep, the dashboard isn't super optimized for very large data sets. I think realistically, it will never really be the best place for very large data sets. Though I agree locking down the UI is not ideal.
I think it handles large number of documents sort-of-ok, but perhaps it's the number of fields that's really breaking it down. The table is virtualized vertically, but not horizontally. That's something we should definitely look into.
Gotcha, that makes sense
Another data point here, I have a table with 20 rows and 14 fields, and no values that aren't very very short (20 chars or less). It takes ~10s to load this table. Might be more than horizontal virtualization.
Like that should be instant with no virtualization
Again, not a huge deal, but I do wonder if it looks bad to folks that are evaluating, as it isn't clear to the observer that it's strictly a client side issue.