This article mainly introduces the multi-person video conferencing Service end architecture, the article from the blog Park Rtc.blacker, reproduced please explain the source.
With the rapid development of mobile Internet, many companies want to intervene in online education, smart home, multi-person video, security monitoring and other fields, although they are video communications, but their service-side architecture and point-to-point communication big do not want the same,
In most cases, single-player video calls may not need to use streaming services at all, and multi-person video, online education, these must be used, so the following mainly introduces the service-side architecture pattern in multiplayer video, and their respective features:
One, mesh structure.
This is the simplest multi-person video Call architecture mode, all media streams do not need to go through the service side, the client directly to peer, you can build multiple peerconnection through WEBRTC, the structure diagram is as follows:
Advantages of the program:
1, the service side pressure is minimal, in most cases do not need to use streaming media services.
Disadvantages of the program:
2, the client load is too large, non-issue expansion, especially mobile, codec pressure will be very large.
Two, mixer structure:
Video conferencing is basically a kind of structure, his biggest feature is the service has done a lot of things, including transcoding, mixing, stacked screen, so the server load is very large, the structure is as follows:
Advantages of the program:
1.1, the client load is minimal, and the same as one load, so theoretically can support a lot of people at the same time video.
1.2, because the server has to do codec, so it can be seamlessly integrated with existing products.
1.3, can maximize the use of hardware capabilities, such as hardware MCU, chip.
Disadvantages of the program:
1.1, the service side load is very big, the construction cost is very high.
1.2, latency problem, because the server does a lot of action, so it will bring a delay.
Three, router structure
The biggest feature of the scheme is that the server is responsible for packet forwarding only, not responsible for transcoding, YY streaming media Services is basically this function, the structure is as follows.
Advantages of the program:
1.1, compared with mixer, the service-side pressure is relatively small. and easy to expand.
1.2, low latency, especially in combination with SVC, can greatly improve the client experience.
Disadvantages of the program:
1.1, considering that different clients need different receiving ability, so the architecture of the server is really not simple.
Additional notes:
1, because the various models have their own advantages and disadvantages, so in practical applications you should choose according to their specific business model of the appropriate architecture.
2, there are a lot of open source frameworks and PAAs can do these things, but the actual application of such problems, immature.
Android IOS WebRTC Audio Video Development Summary (22)