Forum Academy Marketplace Showcase Pricing Features

A fair and honest chat about Performance

I mean, my project is that I’m trying to improve Bubble and fix its issues in a general way that drives product development forward. I get interested in an issue and then just fix it.

For example: There is no general iteration in Bubble. (There is, in fact, iteration in specific all over Bubble, but as Bubble programmers we have limited access to it.) I fixed that (in not just 1, but 3 ways now) via List Shifter. (I’m shortcutting the story here: I introduced that first in Calendar Grid Pro but decided that the techniques were fundamentally important enough that they should be generally available to anyone, even if they were not willing to pay for it. It’s that important.)

Things like List Shifter (and Calendar Grid Pro, in a more specialized sense) are my contribution to making Bubble more performant. There ARE areas in Bubble that are poorly performing, but those areas are NOT what most Bubble thrashers are running into. Mostly, those folks are running into bad design choices on their own part. And, in those cases, the “problems” have little or nothing to do with Bubble (or any other framework) – the thing they’ve built would perform poorly regardless.

Sometimes, these folks are just running into a “you can’t do that in Bubble”. Which is not a performance problem per se, it’s just a thing you can’t do. That’s very different from something that’s “slow”, right? (“Impossible” and “slow” are orthogonal. Something that cannot be done has no speed. But Bubble’s a product that, in part, is aimed at an audience that has no idea of how web things are built. It does a pretty remarkable job of enabling those people to build functional web applications… at the expense of enlightening them not one bit as to what they are actually doing. Go figure.

NOW, the above being said: There are some very strange ways in which Bubble is slow. These are not just limited to the backend, but also involve reading data in the front end. I feel that these are bugs and, should they be properly understood, can be fixed. But I’m gonna tell you that there are very few people that have actually identified where those issues lie.

Most “performance” issues that people report are just “doing it wrong”.

4 Likes

@morgan Great video! Performance wise I have workflows exactly as Keith describes them, and they are fast.

In case you don’t want to copy data from your product to your product order, what you need to do instead is to have a product revision. That is, you have a product (apple) that when created in the database has a revision no. 0 and a status field of isActive = yes. Whenever you change the price or other values on that product, you make a new revision and set the old revision to isActive = no white revision no. 1 isActive = yes.

That way each order connects to the correct product revision, e.g. “apple revision no. 8”.

Of course if you hard delete that item on the database it will get removed at the customers order, no way to circumvent that, but that’s your responsibility to make sure it wont happen.

Hope that helps.

1 Like

Hey Keith, so I’ve watched your video on list popper twice and will give it a go tomorrow, thank you.
I do have a couple of questions and sorry If I missed it in something you’ve written earlier but it’s late here!

  1. I noticed you had a product list of type products. So can I add my products along with all its fields including images etc to a new item via list popper?

  2. What is the main advantage of doing it this way over adding line items via bubble and a recursive API?

I think I gave basically the same answer in the other thread :grinning:

These “Bubble so slow, just a hobbyist toy” rants are almost always bad design. Then the next bubbler who is frustrated finds it and gets discouraged from using a tool that can be literally career changing.

An aside, @morgan, Backend Workflows are not the best option for things that are part of the user experience. Keep it on the front end and your workflows will get immediate priority.

6 Likes

Well put @eli!

2 Likes

I can understand your concerns and I am absolutely persisting with some very generous help from the tight-knit and passionate bubble community. However your comment assumes that my comments are based on bad design or perhaps inexperience of the platform. I’ve built many enterprise database apps and ecommerce websites and I have a pretty good grip on what is good and bad (or inefficient) design. I’m running into some issues because there is a bubble way which in some cases makes things more inefficient. The issues regarding speed I’ve alluded to are very simple Server Side workflows such as copying a few rows of data into another table. It’s painfully slow (even when tested in complete isolation) and I’ve definitely not experienced such speed issues on other platforms using a lot more data such as quickbase (for example). My concerns regarding speed is that if I have to put a 2-5 second gap in between the simplest (API) workflow actions so that bubble doesn’t max out, then what will it be like with thousands of transactions? The answer is illustrated here by someone that uses (or was using) bubble quite extensively on a similar scale to what I am doing (and it’s not a rant, just facts from someone with much more experience than I):

I am not trying to sledge or discourage people from using this amazing and pioneering platform, I want bubble to grow and flourish! I also want bubble to put some more effort into making it more efficient on the backend and to be more transparent about what is currently causing the issues and what will be done to fix it so myself and other users can have confidence in the future of the platform. I have been advised to get a dedicated server but that looks like it may not be the answer either.

It is a fact that bubble is extremely slow on backend server actions regardless of the plan they are on. Sure some people could design their apps more efficiently but that is not dealing with the root cause of issues experienced by many others.

8 Likes

@morgan well said and yes it is the biggest issue and I am sure bubble are making this their main priority at the moment, and before anyone chimes with “wise words of wisdom” no I’m not a Bubble newby.

1 Like

Update: Everything rebuilt via real code, we are super fast!
Cost for AWS is only ~$120/m (I paid thousands / month for bubble dedicated)

I’d still recommend using bubble for testing the waters but be ready to transition to real code when you see success and can afford.

I’m still using bubble for testing new ideas, and even to build prototypes that I can show my developers to duplicate via code, makes explaining ideas much better that I can actually build it as envisioned and simply ask coders to clone my bubble creation.

Also, efficient bubble design and data structure can improve speed but even with the most efficient design - you’ll never be as fast as any app built via code.

Bubble did improve a lot over the years, I consider the addition of the plugin marketplace the best thing that happened.

With that said, I don’t see much improvement in speed. A couple weeks ago I remember bubble being down for almost a full day.

I wish Bubble spent most or all of their efforts on one thing: SPEED, SPEED, SPEED.

Nothing else to improve on bubble before speed issue is resolved.

If improving speed is not possible because of the underlying technology, then that should be clearly communicated to us. But who knows, maybe they are rebuilding the underlying technology right now so they can solve the speed issue.

We might just get a surprise speed improvement before the cost increase goes into effect for all currently on legacy plans.

Once speed is resolved, the future is limitless for Bubble and Bubblers.

11 Likes

How much has this costed?

3 Likes

I’m wondering the exact same question. Bubble cost you $1,000s per month. How much did you pay programmers to remake this for you?

@feee Great post! I do the same… Just finished another MVP in Bubble that will be developed by our internal software team. It saves thousands in programmer cost when you lay it all out for them. More importantly, our clients get to play with the Bubble MVP and make changes BEFORE the costly software team goes in and lays the concrete for the final application. For me, Bubble has been worth its weight in gold.

One item to note that’s most important for our company is uptime. We’ve had 0 outages with Amazon/Digital Ocean over 2020 (so far… knock on wood) and my Bubble MVP has been down 3 or 4 times in 2020 - To be fair, the Bubble outage has been short periods of time and barely noticeable… :slight_smile: But it has to be mentioned.

4 Likes

Wasn’t even that much actually, I now have a team of coders - payroll is the cost.
I can give you a quote for turning your bubble app into real code - maybe I’ll start a side hustle around that :face_with_hand_over_mouth:

3 Likes

Oh good! So at least USD $20,000.00/month in the US then! Thanks for sharing!
I think it’s funny how you say “real code” :joy: like Bubble is surreal code or something like that.
I understand what you mean with it, just think it funny.

6 Likes

I am fairly happy with bubble speed in my use cases.
Can’t say how it would handle 300k users simultaneously because my apps are not in this situation.
If you don’t have this problem, I really don’t see any reason to change from bubble to a bunch of developers.
In my opinion, the best thing about bubble is the ability to deploy new features quick and easy. And INDEPENDENTLY.
If you deliver value in your app, a few miliseconds really won’t mean anything.

4 Likes

Hello, new user here. I came to Bubble in hopes of building a custom work-management tool. After using Airtable for our work-management for several years, we became frustrated with it because Airtable keeps refusing to add several crucial features (link to other bases, advanced permissions, etc).

Then we went to Coda, and it actually has almost all the features that we lacked in Airtable (minus uploading attachments into a record, but that’s another story). However, after investing hundreds of hours building our own app in Coda, we found out it became unusable with even a small number of records. Performance was very slow, and it was not scalable at all. It’s also not a priority for them at the moment (they’re focusing on design and UI improvements among others).

Which is why I came to Bubble. Now the first thing I actually did after playing around in a bit, is see how scalable this is. It’s disheartening to see that performance will become an issue here as well. Hence, I’m not going to repeat what I did with Coda. I won’t spend months on this product, only to find it become unusable, again. So sadly, I won’t be using Bubble

PS: wanted to mention that I’m not looking to make a very complex work-management process. Just a simple task list with a customized workflow, which also links to my small team’s attendance and HR info. I realize that there may be apps built for this, and I’ve searched a lot but none of them fit our needs fully so far, which is why I’ve tried to make a custom one in Airtable. I also realize that it’s not an easy task, if it were that easy to make project management apps, everyone
would’ve done so. So maybe Bubble isn’t meant for this, but seeing how close Airtable got (their performance for my use case is perfect, but it lacks customized user permissions, automations and buttons), I wish Coda/Bubble were doing it.

Hi JarvisDigital

Thanks for sharing your insight.

Just for my understanding, and hopefully useful to others in this forum too, could you please share some light on what “tests” you ran and felt that “bubble subsequently failed on”, in order to come to the following conclusion:

It’s disheartening to see that performance will become an issue

Many thanks, and wish you all the best in your application!

2 Likes

@allenyang, I just finished reading the book on JAMstack (written by its creators). From what I can tell, this design approach may be the next best way of making apps. Between the experience of my developer friends and various case studies
LobLaw:

  • 9.4x Faster To Interactive
  • 92% Better performance
  • 38k monthly cost savings

Smashing Magazine:

“The time to first load is so much faster than before… before we had to wait for the HTML file being served for 800ms and now it’s 80ms .” (more here)

and the fact that:

more than 8 percent of the internet population visits a [*JAMstack] site each month

this looks like the future of web development. Perhaps JAMstack itself isn’t the future but the central tenant of improving performance by pre-rendering HTML (meanwhile getting any CRUD operations done via APIs) is .

I know that Bubble already uses React on the front end.
You guys have also separated the concerns of the front end from the back end.
From the little I know about your tech stack, it seems like it has the essential structures in place for pre-rendering HTML like JAMstack would. Is this something you guys are experimenting with (I imagine the answer is yes)? Can you tell us the results of those tests? Or are performance issues mainly from other areas of the application?

4 Likes

I’m confused, the forums said this is your first post. Do you have other accounts that you have used to get help, and tips to optimize your app? If not, I’m pretty sure the design is likely the issue. If you have lots of searches and awkward groups, that would cause slowness. Additionally you didn’t actually say what any of your issues are.

You only spent a couple hundred hours on another app design platform, so now you won’t spend much time here, but then come on the forum and suggest you really know the performance of the platform? I would suggest people should not throw stones in glass houses that they aren’t familiar with.

Now if you want to show us your editor, confirm plan type (for speed and capacity), and tell us the issues you are having, then a meaningful conversation can be had. If this isn’t done, I would strongly suggest this post be taken with a grain of salt.

While it sounds like I’m bashing you, I’m really just trying to ensure everyone has the most optimal time possible. If you would like the community to ensure that the performance is the issue, please show us your project so we can help you. Also keep in mind that the free plan is in itself slower, so if you are on that, it will be half the battle.

3 Likes

I just put up a post on the performance issue. Wondering about other people’s recent experiences.

Are you still on bubble?

1 Like