37

Kranky Geek WebRTC 2015 - Beyond P2P: Video routing in WebRTC

Embed Size (px)

Citation preview

we love WebRTC because it is p2p

better security lower latency less hassle lower cost

BS METER THEY SAY P2P HAS: 90% BS

25%

75%

25%

CONFERENCING RECORDING TALKING TO PHONES BROADCASTING

RECORDING

QUICK CONNECT

NAT TRAVERSAL

THE MCU THE SFU P2P MESH

THE MCU THE SFU P2P MESH

MIX

THE MCU THE SFU P2P MESS

MIX

C

ALL ABOUT THE PERF

A L L A B O U T INTEROP

ALL ABOUT … NOT SURE

THE SFU

ALL ABOUT THE PERF

20% CPU

1000 video streams 550

Megabits https://jitsi.org/videobridge/

performance

WORK FOR A MOBILE?

an MCU only sends 1 stream

an SFU needs tons of bandwith but … from a participant perspective

WOULD AN SFU EVER

SIM & SVC in a

nutshell

s i m u l c a s t

SVC: SCALABLE VIDEO CODING

SVC: SCALABLE VIDEO CODING

SVC: SCALABLE VIDEO CODING

SVC: SCALABLE VIDEO CODING

SVC: SCALABLE VIDEO CODING

SPACIAL SVC

TEMPORAL SVC

simulcast in chrome o=- 3407482598521328076 2 IN IP4 127.0.0.1

s=- t=0 0 a=group:BUNDLE audio video data ... m=audio 9 RTP/SAVPF 111 103 104 9 0 8 106 105 13 126 ... m=video 9 RTP/SAVPF 100 116 117 96 ... cr

eate

Of

fer

simulcast in chrome o=- 3407482598521328076 2 IN IP4 127.0.0.1

s=- t=0 0 ... m=audio 9 RTP/SAVPF 111 103 104 9 0 8 106 105 13 126 ... m=video 9 RTP/SAVPF 100 116 117 96 a=ssrc-group:SIM 3535700627 3169665807 0902838992 a=ssrc:3535700627 cname:localCname a=ssrc:3535700627 msid:DN4NQ 0a1622a6a67d a=ssrc:3169665807 cname:localCname a=ssrc:3169665807 msid:DN4NQ 0a1622a6a67d a=ssrc:0902838992 cname:localCname a=ssrc:0902838992 msid:DN4NQ 0a1622a6a67d setL

ocal

Des

crip

tion

sim

ulc

ast

simulcast: raison d'être

360p

720p

1080p

using endpoint generated bandwidth estimations (REMB) draft-alvestrand-rmcat-congestion draft-alvestrand-rmcat-remb

simulcast: raison d'être

360p

720p

1080p

using endpoint generated bandwidth estimations (REMB) draft-alvestrand-rmcat-congestion draft-alvestrand-rmcat-remb

1080p

simulcast: raison d'être

360p

720p

1080p

using endpoint generated bandwidth estimations (REMB) draft-alvestrand-rmcat-congestion draft-alvestrand-rmcat-remb

1080p

fiber or cable

simulcast: raison d'être

360p

720p

1080p

using endpoint generated bandwidth estimations (REMB) draft-alvestrand-rmcat-congestion draft-alvestrand-rmcat-remb

Wi-Fi

720p

simulcast: raison d'être

360p

720p

1080p

using endpoint generated bandwidth estimations (REMB) draft-alvestrand-rmcat-congestion draft-alvestrand-rmcat-remb

3g

360p

simulcast: raison d'être

360p

720p

1080p

using endpoint generated bandwidth estimations (REMB) draft-alvestrand-rmcat-congestion draft-alvestrand-rmcat-remb

3g

360p

fiber or cable

simulcast: raison d'être

360p

720p

1080p

using endpoint generated bandwidth estimations (REMB) draft-alvestrand-rmcat-congestion draft-alvestrand-rmcat-remb

3g

360p

Wi-Fi

simulcast: raison d'être

360p

720p

using endpoint generated bandwidth estimations (REMB) draft-alvestrand-rmcat-congestion draft-alvestrand-rmcat-remb

360p

3g

360p

3g

no yes no

no yes no

soon yes soon

How about the W3C and IETF ?

(to be filled in on Thursday)

20% CPU

1000 video streams 550

Megabits