Help with techstack for multi-user livestreaming social app

[pearlgate]

Hi, we are two GenZ cofounders (24 and 21) building the first study stream platform in Southeast Asia to help young professionals focus and feel less lonely. Check us out at: www.lean.social

Our biggest struggle right now is our video streaming infrastructure and its prohibitive cost. We are using P2P WebRTC and each video room has a limit of 5 participants. There is no cost but it's very limiting and low-performance.

If we wanted more, we have 2 options:

1/ Buy a video sdk server plan (Twilio, Daily.co, Agora, Amazon Chime, etc):

- The cheapest these can get is 0.0015$ per participant minute. Without scale it is 0.004$/participant minute

- This is prohibitively expensive for us with limited funding. 1000 users using our service for 2 hours daily is a 10$k bill montly

Is there any more affordable providers? How much of the cost can we save if:

- We increase the latency to 10-15s? Real-time is not important for us

- We Reduce the video resolution to super low ~ 240p? Our video frame is very small to increase focus

2/ Build our own self-hosted server:

- We are leaning towards open-source projects like LiveKit (seems better) and Jitsi. We plan to host the instances on AWS because of our CTO's familiarity with its infrastructure.

Has anyone tried building B2C social apps with video calls from open-source projects like these? Could you share:

- How much does self-hosting on aws would save us?

- How deeply can we customize their UI? We don't do anything like the usual video conferencing layout. We want to build our own layout which is vastly different.

 

[613jono]

@pearlgate I think it’s a strategy and priority choice you have here.

Sure you don’t have money but that’s honestly not an issue. If you have 1000 users using 2 hours daily, then money is not an issue I can tell you that. It’s easy to make money out of this KPI. Business model OR startup funding will do the jobs. And 1000 users active daily is way more than enough to find this money

Now there is the product choice. Is quality important ? Interoperability ? Speed ? Price ? Really depends.

If this was me, I would start on Amazon Chime. Low quality, easy to setup, one of the cheapest. And the real upside here is that you can get those AWS credits really easily, it will absorb costs for a while. Will let you go forward for some time until you get clients.

I think the priority is making product work reliably and prove that there is a market.
Then I am 100% sure by the time you get a lot of active users, you’ll either meet people with experience or money, or you’ll find another solution.

PS.: hosting it yourself is a bad idea for me. Way too hard to do properly, and the economy at your level is really low. And even it will cost money since Chime is on demande and you ll paye what you use. So no user no cost

 

[jsigler]

@pearlgate (sorry for my english)

I create an app using MediaSoup in Ionic React. The documentation ir poor. But they have demo app for clone in github.

MediaSoup works form MCU arquiteture, you can have easily 50 participants in same room.

In specific event, my app supported 90 participants in same room.

Our app recibe 5K user per day. Sometime we have 20 rooms with 4/8 people simultaneos.