Chat about hosting and infrastructure decision
I truly don't understand adding Pulumi when you've gone out of your way to essentially avoid managing your own infrastructure.
28 Replies
Yeah, that is also my main concern right now. These are merely considerations at this point, and I’m just trying to think ahead in terms of how to maintain the project’s scalability
Ask web dev Cody, he has experience with both aws amplify and convex
Yeah, a video he posted about 1 year ago i think, did pop up in regards to amplify. Erquhart is very right about the whole not being invested in aws already or the other services in amplify. Ive had some thoughts about this too.
Yeah I think he moved an app from amplify to SST
We're the first wave of product engineers to learn about Convex and take it seriously - it's a competitive advantage. I think we'll find a lot of traditional recommendations won't take into account what a game changer it is to have the vast majority of your backend needs handled and scaled by a single service. This will change as the industry catches up. I also expect folks a lot smarter and more experienced than I am to make the wrong recommendations, because no one has pulled off the moonshot that Convex is on track to achieve.
At a glance, many tenured engineers and architects will not expect this to be a serious, scalable production solution. As recently as a year ago I was one of them. Attempts like Firebase and Meteor taught everyone this lesson. Again, I expect this will absolutely change once startups using Convex begin noticeably outpacing their competitors and the industry begins to take notice.
I too discounted convex too early. I thought, what, no sql? I’m good at sql. But I’m facing the realities of building realtime and cache myself. So I’m back haha. But anyway, I think this thread started with hosting and it is an interesting question as to where to host the front end. For example, There’s only a couple ways to easily host next.js
@erquhart - do you have an affiliation to netlify?
Used to work there
To your example about firebase, I think they actually solved a lot of problems in their day. I built a prototype on Firebase last year. Still good today, but I get the sense that firebase isn’t prioritizing the react ecosystem and is instead prioritizing gcp, especially cloud run
Like they haven’t updated their auth client libraries for a years. Framework aware hosting was still beta last year.
Im hoping convex’s open sourcing efforts can alleviate some concerns that an acquisition of Convex Inc would cause the product to go into maintenance mode
For sure, Firebase was compelling and had some good road ahead of it until the Google acquisition.
It was still the best way to spin up an mvp prior to Convex. But it's also well understood to be a solution that does not scale very far beyond mvp.
That's the barrier for Convex to break.
Thanks a lot. Very valuable information and thoughts being shared. Will be steering away from aws for now.
Still dont know what opinion to make of Netlify, based on some previous incidents regarding pricing, that came up when searching the web.
I will take a closer look at Netlify and Cloudflare, and gain some more insights to base my opinion of.
I know one recent and viral situation that probably came up. imo it was mostly sensational, Netlify came out really clear on the mistake on their side that led to that, the dev was never in danger of being required to pay. That said, I'd rather see you go to Vercel than something that adds more complexity to your frontend.
Yeah you make a valid point.
@oscklm - so is your front end Next.js? Looking around, I only see that Netlify and Vercel makes it easy to host next.js. SST says that they show you how AWS works, but I've never tried it myself. It would be helpful for the next.js community if more people tried options aside from Netlify and Vercel. The thing is, Next.js moves so fast that it's hard for companies aside from Vercel to accommodate corner cases with each Next.js release.
Expo for native and next js for web for an internal admin dashboard yeah, and we are currently using vercel, and its a hate n love relationship for me. And honestly not at all looking to change that in terms of where at now. Im kinda just sniffing out the possibilities and gaining some deeper understanding of alternatives, for the future in terms of scalability.
And yeah i agree with you on that, i also find what they are doing with SST really interesting. Have decided to not really dive too much into SST until they are done with their rather big transition they are currently making (as stated on their website) but i wanna try it out on a personal project sometime
whats the scalability issue? technical, pricing, or something else?
The scalability issue is primarily related to pricing. Mainly centered around a wish to transition to self hosting our video infrastructure and ideally an identity provider. To reduce our dependency on services like Mux and Clerk.
Did Vercels pricing structure change last week help?
Pardon? I don’t believe I’ve mentioned anything regarding vercel pricing. If i have that was a mistake. But i do wish to explore alternatives to vercel.
You said you're on vercel and the scalability concern was around pricing
assuming that was the correlation
I see. Just wanted to state that we were using vercel, in response to question if our front end was next js. Sorry for the confusion.
The concern with scalabilty/pricing is clerk and mux
Wow, do you really have a cost issue with Clerk for an internal app? Do you work at a huge company (and one that doesn't share user accounts across Clerk)?
I wanna make it clear that I’m not / company i work for is not having any issues as of right now. These are mostly some of my own concerns. No huge company. A small start up, trying to build a video platform.
Id just like to explore the alternatives and try to think a bit ahead, mostly out curiosity and wanting to gain a bit more knowledge on the subject of infrastructure.
Clerk pricing or scalability is def not a concern for our internal admin dashboard. But our platform consists of more than that.
Oh, I see you are talking about Clerk's costs on the end users of your platform
Yeah, it's a concern. Clerk is just really well positioned right now for the React ecosystem
Yeah exactly. Thats the part thats a growing concern for me, as we hopefully reach a point where that will be a real life problem 😂
I am praying too haha. Clerk lowered prices last year.
Yeah def, i don’t see myself being much successful in finding any better alternatives to be honest
I don’t know why, but part of me tends to go off on a tangent with myself, trying to optimize things that obviously haven’t become issues yet. But I can’t help being concerned about whether I’ve made the right choices.
My thoughts is to just move forward. It's not really something you can control. Auth0 went from the SMB market and then raised prices for enterprise. Clerk may do the same. Or, they might go into maintenance mode like Firebase auth
Keep to social auth - that would be easier to migrate than passwords
I appricrate it. And im sure your right about that. Thanks for the convo man
Yeah valid point