Thanks for listening to the community feedback.
Look forward to seeing how this effects my WU usage once the new calculations are in play.
2 Likes
was the problems resolved? Can I start producing again?
This is so awesome and so fair. We look forward to trying this out later this week. Well done @josh
Omar
2 Likes
Putting on my mathematician hat for a moment, yes I have a graduate degree in Mathematical Statistics. I can tell you regression is an invalid methodology for developing your pricing model. It is well documented that memory and time consumption in computing is governed by “power law”, also know as “fat-tail”, or “black-swan” distributions. These stochastic processes violate the assumptions that underpin regression models.
The theory behind this is that computing resource consumption is a multiplicative, not additive process and so leads to “log-normal” types of distributions, instead of the standard “Gaussian” distributions of classical regression theory.
The upshot is that the you are far more likely to have applications that incur unexpectedly high bills for their design, than the simple averages and standard deviations of regression models would predict.
61 Likes
This is a great step in the right direction. I think you answered almost all of my questions. I look forward to continual tweaks and more tools to help weed out bad actors (such as capping the max units a user is able to trigger), and viewing the cost associated with each user.
I will say that this is the second time in two years that my livelihood has been questioned because of a pricing update – it is hard to continually trust a platform and leadership that does this to you more than once. I hope that you invest some more time and money into improving your connection and collaboration with the community.
I think you spoke to it a little, but I think an explanation to @aaronsheldon’s point as to why bulk processing thousands of data points on AWS does not meet the minimum thresholding on building, but would cost thousands of dollars on Bubble is needed.
23 Likes
Double that - Bubble always get it right in the end 
but they do love some drama 
1 Like
I get it. But is it not weird that we almost must make a phd study out of it before it even makes sense?
And what to think about customer discussion. Apart from requirement analysis we also have to do a wild guess as to how much WU the to build application will cost. Which can vary from anything between let’s say $30 to $3000 a month.
I am baffled that smart people come up which such a complex pricing strategy that of course no one in the market will understand.
Who will benefit from this? Nobody. And that’s sad to conclude.
23 Likes
aaron bro i always love your posts (super helpful) but it makes me feel like my english isn’t good enough
10 Likes
He’s just saying that they calculate their metrics in a fundamentally wrong way. And basically he’s right.
15 Likes
@josh @emmanuel Thanks for the update. Appreciated that the feedback was heard and you are adding transparency to the metrics for our own planning. I will still evaluate what the pricing and workload will be post-changes, but it is definitely a step in the right direction.
I have no clue what he said. English is my first and only language. I will follow that man anywhere blindly.
12 Likes
his statement confused me more than @josh post ,lol
2 Likes
Looking at the usage metrics for my app, one of the biggest consumers of WUs (at least under the old weights) is being driven by the following piece of code:
All I’m trying to achieve with this piece of code is to present each user with a count of the documents they have stored in my app – it just shows up on the main app dashboard as a text (e.g. “325”) – so it is surprising to see such a simple action consuming such a large portion of my app’s WUs.
Is there a way to better optimize this? I’m guessing this is using so many WUs because behind the scenes Bubble has to search the database for ALL documents and returns a lot of data when all I really want is a simple count.
Is there no way to have the server/database perform the count and return that information instead of returning all details of all database records and then doing a count action on the client side? As we move to this new workload-based paradigm it seems that Bubble needs to make improvements to its systems to help us better optimize our apps.
15 Likes
Thanks @josh, @emmanuel and Bubble team for this update. Things got wild lately, I hope we can resolve these issues.
Awesome, excited to see these changes and results. WUs are the way to go, we just need to get quotas, pricing and predictability right.
Great news!
Last week wasn’t great for anyone around here in the community and I imagine it wasn’t an easy week out there either, but we’ll get through it.
Thanks again Josh, Emannuel and team.
1 Like
But I don´t understand, they updated the pricing but are we getting any better performance, EU servers, better do a search for´s, …?
Or is just hey, Bubble does not have any upgrade but you´re going to pay more. That´s it?
6 Likes
Thanks for the resposne. I’m eager to see the improvement in the WU count on friday to better understand the impact
People will be appeased because they’ll (hopefully) see their WU usage decrease significantly tomorrow.
But the core issue still remains that WU is arbitrary and could change at any moment.
And I’m still estimating at least a 10x increase in cost using the new calculator. Nothing like the “<$250 increase seen by 88% of apps”.
I still can’t fathom where your numbers came from for those estimates or for calculating WUs and that’s problematic for me.
18 Likes
Just wanted say RIP to server side plugin actions. With them costing 0.2 WU and another 0.0005 WU per millisecond (holy $hit that is expensive !!!) to run, it’s safe to say running they are officially dead since it normally takes about 3 seconds just for the action to start when it’s “cold”
RIP Wait Please and others!
28 Likes
@josh So, what about frontend actions? (Like execution javascript, set state, reload page)
8 Likes
Oh man JIT (“Just-in-time”) warm-up of server-side actions is going to kill us all.
14 Likes