Skip to content
This repository has been archived by the owner on Nov 11, 2024. It is now read-only.

CDN Transcode Sample Getting Started Guide

Jump to bottom
Zhizhen Tang edited this page Jan 14, 2020 · 49 revisions

Table of Contents

This document describes how to run the CDN Transcode Sample step by step.

The sample provides two kinds of services - live streaming and VOD, and this guide shows how to use the services in a simplest and typical way which can be scaled out to more complex environment. E.g.: in this guide, the docker images for transcoder server and cdn edge server are hosted on the same physical server. In real case, they can be hosted on different servers located in different places in the CDN network.

Architecture

The sample implements a reference server-side transcode system over CDN infrastructure, which features live streaming and VOD.

The workflow of live streaming is as follows:

  • The Live transcode Service receives the video stream locally or over rtmp, decapsulate and demux the flv video, transcode to other codecs/bitrate/resolution, in 1:N manner, which means one input and N output. E.g.: one channel HEVC 4k@60fps, transcoded into one channle AVC 720p@30fps, one channel HEVC 1080p@30fps and one channel HEVC 4k@60fps. The transcoded video streams are then be encapsulated in flv over rtmp, be distributed to the CDN service.
  • The CDN service receives the video streams from the Live transcode Service, and will translate the video streams into HLS/DASH format.
  • The client starts playback the video.

The workflow of VOD is as follows:

  • The client starts playback of a video by requesting the video manifest file.
  • The CDN service checks if the video has been transcoded. If the DASH/HLS segments exist, the CDN service simply returns the manifest (and any subsequent video segments.) Otherwise, the CDN service schedules transcoding of the video in Kafka, a message system shared between the CDN service and the VOD trancode service.
  • The VOD trancode service receives the request and starts transcoding of the video clip.
  • The CDN service returns the resulting DASH/HLS manifest and segments to the client.

Prerequisites

In this document, we'll use the simplest example to show how to build up the pipeline for different user scenarios. To simply the setup, we'll host docker nodes on the same physical server (named as "CDN-Transcode Server" in this document). One Streaming Server will be used as well to RTMP stream the source video content to CDN-Transcode Server, however you can omit Streaming Server if you want to just use local video content on CDN-Transcode Server. A client system is also needed to playback the transcoded video streams.

Below is the basic block diagram for the sample setup:

 +------------------+   +----------------------+   +------------+  
 |                  |   |                      |   |            |  
 | Streaming Server +---+ CDN-Transcode Server +---+   Client   |  
 |                  |   |                      |   |            |  
 +------------------+   +----------------------+   +------------+

Note: for live streaming and VOD in this document, it refers the interaction between CDN-Transcode Server and Client, not refers interaction between Streaming Server and CDN-Transcode Server.

Setup CDN-Transcode Server

  • Install Ubuntu 18.04/CentOS 7.6 on CDN-Transcode Server, and configure the IP address & proxy properlly.
  • Install the third-party dependency Libraries and tools.
script/install_dependency.sh

Note: This script must be run as root.

  • Setup docker network proxy
    You may need to setup the docker proxy on CDN-Transcode Server. Below is an example to directly use the host proxy as the docker proxy.
sudo mkdir -p /etc/systemd/system/docker.service.d
printf "[Service]\nEnvironment=\"HTTPS_PROXY=$https_proxy\" \"NO_PROXY=$no_proxy\"\n" | sudo tee /etc/systemd/system/docker.service.d/proxy.conf
sudo systemctl daemon-reload
sudo systemctl restart docker

Setup Streaming Server (Optional)

As mentioned above, Streaing Server is not a must, you can skip this section if you do not want to stream the source via RTMP.

  • Install Ubuntu 18.04 on Streaming Server, and configure the IP address & proxy properlly.
  • Install FFmpeg
sudo apt-get install -y ffmpeg
  • Install Nginx and Nginx HTTP FLV Module
git clone https://github.com/winshining/nginx-http-flv-module.git

wget http://nginx.org/download/nginx-1.15.12.tar.gz
tar -xvzf nginx-1.15.12.tar.gz
cd nginx-1.15.12
./configure --prefix=/usr/local/nginx --conf-path=/etc/nginx/nginx.conf --add-module=/path/to/nginx-http-flv-module
make && sudo make install

Setup Client

In this document, we can use both VLC and web browser as the playback tool. To use VLC, it's recommended to use Windows VLC client. For web brower, in theory any web browser could be OK, like Chrome, Firefox, Microsoft Edge, IOS Safari, however HEVC playback is not well supported by web browsers, so if you want to playback HEVC stream, use VLC. Also AV1 playback is only supported by Firefox.

Build

On CDN-Transcode Server, run below command to build docker images:

mkdir build
cd build
cmake ..
make

Deploy

The sample supports both auto deployment and manual deployment. The auto deployment will help users to have quick try to use this sample while the manual deployment give more flexibility for customerization.

Auto deployment

The auto deploy supports both docker swarm and kubernetes. The auto deploy will deploy the live streaming and VOD services automatically using a default video stream "big buck bunny" as an example. To simply the setup, the original source video content are local content but not streamed from the RTMP streaming server.

Note: If you want to use other video clips to try the auto deployment, you can simply visit https://<CDN-Transcode Server IP address>/ using web browser, click the guest button at the top right-hand corner and then choose upload option. It's recommended to use .mp4 video clip files.

Auto deployment using docker swarm

For details, please refer to Deploy on Xeon E5 using docker swarm

Auto deployment using Kubernetes

  1. Setup kubernetes environment For details, please refer to Setup Kubernetes environment
    Note: If you have setup kubernetes environment, please skip this step.
  2. Setup Kubernetes logging & monitoring environment For details, please refer to Setup Kubernetes logging & monitoring environment
  3. Auto deployment on your own platform

Manual deployment

The manual deployment give an example to users who want to use this sample as a base to build up more complicated CDN transcode solution. For details, please refer to Manual Deploy

Powered by Open Visual Cloud media software stack.

Design

Setup

Deployment

BKM

Remove

Example

Tutorial

Clone this wiki locally