https://blog.discord.com/how-discord-handles-two-and-half-million-concurrent-voice-users-using-webrtc-ce01c3187429
From the very start, we made very conscious engineering and product decisions to keep Discord well suited for voice chat while playing your favorite game with your friends. These decisions enabled us to massively scale our operation with a small team and limited resources.
This post gives a brief overview of the different technologies Discord uses to make audio/video communications a seamless reality.
For clarity, we will use the term “guild” to represent a collection of users and channels — they are called “servers” in the client. The term “server” will instead be used here to describe our backend infrastructure.
Every audio/video communication in Discord is multiparty. Supporting large group channels (we have seen 1000 people taking turns speaking) requires client-server networking architecture because peer-to-peer networking becomes prohibitively expensive as the number of participants increases.
Routing all your network traffic through Discord servers also ensures that your IP address is never leaked whether you use text, voice, or video — preventing anyone from finding out your IP address and launching a DDoS attack against you. Routing audio/video through media servers offers other advantages as well, such as moderation. For example, administrators can disable audio/video for offending participants.
Discord runs on lots of platforms.
The only way our team can support all these platforms is to take advantage of code re-use and WebRTC. WebRTC is a specification for real-time communication comprised of networking, audio, and video components standardized by both World Wide Web Consortium and Internet Engineering Task Force. WebRTC is available in all modern browsers and also as a native library to embed into applications.
Discord’s audio and video features are implemented using WebRTC. This means our browser app relies on the WebRTC implementation offered by the browser. Our desktop, iOS, and Android applications, however, make use of a single C++ media engine built on top of the WebRTC native library — specifically tailored to the needs of our users. This means that certain features work better in the installed application than in the browser. For example, in our native apps we can: