Professional Support?

Job Opportunities 💰
1

Does the MediaSoup team offer professional support?

We can’t find anything with regards here or on the mediasoup website.
We’d like to pay for consulting on our project, and - we’d also like to know support is available if necessary once we are in production.

The first thing we need help with is server setup and allocation. Our client-side application [SPA] is running on separate servers. We want our SPA to connect to a separate mediasoup server by IP only (no domain/ssl/etc), and have the capacity to serve 50-100 participants in 1080p for 2hrs per show.

Questions:

  1. What is the most cost-effective cloud service provider for our needs (AWS, DigitalOcean, Linode)?
  2. What is the most optimal server setup for our needs (OS, CPU, RAM, HD, NET, etc)?
  3. What is the ‘best-practice’ installation process for this mediasoup server (server software/node packages, firewall config, networking, mediasoup package installation, run-scripts).

Capacity Requirements:
Each instance must support 8 2hr ‘meetings’ per month, with up to 100 participants in full multicast SFU mode. We want 3 different servers setup, with the capacity to serve the same number of show/participants, in higher video resolutions:

Option1: LD + MD + HD
Option2: LD + MD + HD + UHD
Option3: LD + MD + HD + UHD + 4k

(resolution reference):
4k: 3840x2160.
UHD: 1920x1080
HD : 1280x720.
MD: 854x480.
LD: 640x360.

Please contact me directly (by reply/pm etc.) if you can help with any of this.
Thank you

2

Not at this moment. This forum category is for looking people that may give such a support.

3

Thanks for the prompt reply @ibc

Is there a better place to post this request?
Can you provide a list of partners you might recommend?

Cheers

4

I don’t have such a list, sorry.

5

100 participants in full sfu mode, in such resolution??? It is crazy!!!

Each participants need consume video and audio from other 99 participants, it is very difficult to scheduling the consuming behavior use witch core of cpu.

Because the consuming behavior is a CPU-intensive task. As the design of mediasoup, we usually create one worker in one core, and create one router in one worker, then one participant connect one router by send/recv transport . But one core can not support so many consumers, so the pariticipants must be distributed in different router(=>wokrer=>core), even in different server. also we must use some router as Middleware to transport data.

Along with paritipant join, every core in use may reach full. So it is very difficult to scheduling the consuming behavior use witch core of cpu.

I have a distributed solution, but I have not test in such scale.

1 Like