for anyone interested I spent a few hours today testing and logging searches and actions performed by bubble and their WU costs
was rather painful to sift through the reports but I got there
I’d appreciate anyone who wants to chime in with suggestions of things that should be tested and logged into the table. (obviously a crude MVP, but it achieves the desired result of comparing A to B)
essentially I wanted to create a place that I could use as a reference for WU costs in an effort to try and optimize workflows and clarify some of the points made by other people on the forum.
There were some rather unexpected results like paging a searched list that returns 100 (without paging) to 100 items per page reduced the WU slightly - even though the same 100 items are returned in both lists.
search for : count is 0 - this was in a forum thread and originally was being charged around 10 WU in that thread in 2023, the bug was fixed by bubble and it now charges 0.3 WU
returning a list of 10 things stored on a data costs less than a search for the same 10 things using 1 constraint
filtering the current page item’s list of 10 items is roughly 50% cheaper than searching for those same items with search constraints
ie datas>items vs search for items where thing = current page thing
I’d like to run some more tests on larger lists as I believe there will be a point where search is more performant than holding the list on a thing.
please chime in with some suggested tests - I’d love to build this table out and share it. the forum is a real spaghetti mess to sift through to try and find this data and bubble documentation is very limited (and often wrong).
Great idea and a fantastic initiative! To help streamline this and bring more clarity to the questions raised in this post, I’ve reached out to Bubble support, requesting that they interact here directly and share what Bubble is doing to address these issues.
They already got that. Flusk is focused (and does a good job of) ongoing monitoring, but there’s a hole I want to plug with audits of app logic (starting with security, but expanding to performance/user experience).
Basically, the stuff I already do for $1, that I want to make free Bubble AI apps are my area of expertise, so I’m using that to build a tool that will:
identify security issues with context awareness
identify areas of optimisation in the app (audit every single workflow!)
chat with your Bubble app (the logic, not the data)
Hopefully it’ll be useful.
Edit: I’ll make a separate thread at some point soon to avoid clogging up this one
I’ve attempted to perform several more tests today.
I was able to get a successful result in about half of them. The other half are not showing in the workload usage report lol (I’ve waited an hour and they still don’t exist)
** they may be showing but possibly grouped under “individual data request” when I click that nothing shows.
I’m going to trawl through the server logs now to see if I can pick out the missing numbers.
** server logs were no help. they didn’t have any of the missing workflows… actually they hardly showed any workflows at all (maybe half?)
Hi @ppaulorodrigo27 , totally agree, i wrote the first post about data not being deleted, and i can tell you that i has a LOT of exchanges with bubble support…
The support finally deleted data manually…
It’s totally crazy to leave this situation like this without even warning bubble dev as the documentation does not reflect the reality
@fede.bubble would it be possible to create a category in forum for “workload units” or “WU Tips” etc? That way one can track all the new learnings related to WUs that people have shared but are not covered in documentation.
Sure, but “WU Optimisation” makes it much more specific and hence better. In “optimisation” category some other stuff might come too e.g. Optimising the time to build apps. It is like we have a category for “mobile” or “responsiveness”, while we could as well name them “devices” or “Portability”. Specific naming makes it clear.
But if Optimisation is the category name that works better for majority, then let that be it.
While at it you may want to create a “workload bug” category since we are 2 yrs into it and it still seems endless at this point. Most of the user base is either oblivious and being over charged or have came to terms with it.