IVS RTMP Publishing | Real-Time Streaming

This document outlines the process of publishing to an IVS stage using RTMP. For additional details on various ingest options, refer to the Stream Ingest documentation

Prerequisites

Create a Stage

To create a stage, use the following command:

aws ivs-realtime create-stage --name "test-stage"

See CreateStage for details, including the response.

Important: In the response, note the endpoints field, which lists both RTMP and RTMPS endpoints. These are required for setting up your RTMP encoder.

Create an Ingest Configuration

To publish to a stage using RTMPS, you must first create an ingest configuration and associate it with your stage. When you publish to the stage (using the stream key from the ingest configuration and the RTMP endpoint from the stage), the media will be published to the stage as a participant. You have the option to specify a userId and custom attributes, which will be associated with the participant that connects to the stage.

aws ivs-realtime create-ingest-configuration \
  --name 'test' \
  --stage-arn arn:aws:ivs:us-east-1:123456789012:stage/8faHz1SQp0ik \
  --user-id '123' \
  --ingest-protocol 'RTMPS'

See CreateIngestConfiguration for details, including the response.

When creating an ingest configuration, you can associate it with a specific stage ARN up front. Without this association, the stream key is unusable. Also, ingest configurations (including the stageArn field) can be updated via the UpdateIngestConfiguration operation, allowing you to reuse the same configuration for different stages.

Note: The ingest configuration insecureIngest field defaults to false, requiring the use of RTMPS. RTMP connections will be rejected. If you must use RTMP, set insecureIngest to true. We recommend using RTMPS unless you have specific and verified use cases that require RTMP.

RTMP Single-Track Video

Below we describe how to use OBS Studio; however, you can use any RTMP encoder that meets the IVS media specifications.

OBS Guide

Important: If your encoder’s maximum bitrate is set to 8.5 Mbps, the publisher occasionally disappears from the session. This is because the maximum bitrate setting is only a target, and encoders occasionally go over the target. To prevent this, set your encoder’s maximum bitrate lower; e.g. to 6 Mbps.

E-RTMP Multitrack Video

IVS supports the multitrack video capability of E-RTMP (Enhanced Real-Time Messaging Protocol), which allows you to publish multiple video qualities in a single RTMP stream to your IVS stage. This enables adaptive bitrate streaming, so subscribers can automatically watch in the best quality for their network connection.

Once ingested, the different video qualities are delivered to subscribers as simulcast layers. To configure which layers are received by subscribers, see the "Layered Encoding with Simulcast" sections in the real-time streaming broadcast SDK guides: Android, iOS, and Web.

For sample code, see aws-samples/sample-amazon-ivs-multitrack-video on GitHub.

This diagram illustrates how publishing with multitrack video works:

OBS Guide

Publishing with FFmpeg

You can use FFmpeg to publish live video and audio to IVS real-time streaming over RTMP. FFmpeg is a free, open-source project comprising a comprehensive suite of software libraries and tools for processing video, audio, and other multimedia content.

The following example command publishes a stream that includes a color pattern and a tone:

ffmpeg \
 -re \
 -f lavfi -i testsrc=d=300:s=1280x720:r=60,format=yuv420p \
 -f lavfi -i sine=f=440:b=4:d=300 \
 -c:v libx264 \
 -b:v 2500k \
 -g 60 -bf 0 \
 -profile:v baseline \
 -preset veryfast \
 -tune zerolatency \
 -x264opts sliced-threads=0 \
 -c:a aac \
 -ac 2 \
 -b:a 160k \
 -ar 48000 \
 -f flv \
 rtmps://$INGEST_ENDPOINT/app/$STREAM_KEY

In the example, replace $INGEST_ENDPOINT and $STREAM_KEY with your own values from the IVS console or API.

This configuration meets the Supported Media Specifications for IVS real-time streaming, including H.264 video (baseline profile, no B-frames, no sliced threads) and AAC audio.

Private Ingest to Stages

You can publish RTMP(S) and E-RTMP(S) streams to a stage from resources inside your Amazon VPC or from Direct Connect, by using an interface VPC endpoint. This enables a private connection between your VPC and IVS, keeping ingest traffic within the AWS network. To set up and configure an interface VPC endpoint for IVS, see IVS Private Ingest in the IVS Low-Latency Streaming User Guide.

Redundant Ingest

Redundant ingest enables streaming from two separate encoders simultaneously to a single stage, with automated failover for the same source media. This helps protect against source encoder failures and first-mile network issues. Redundant ingest is supported for RTMP(S) and E-RTMP(S) streams.

To enable redundant ingest, set redundantIngest to true when creating your ingest configuration via CreateIngestConfiguration. IVS provides two RTMP stream keys. Configure two separate encoders using the same ingest endpoint with their respective stream keys.

Each physical stream appears as a separate participant in participant APIs (e.g., ListParticipants). However, subscribers can subscribe to only one virtual participant, identified by the top-level participantId in the ingest configuration. IVS automatically controls which physical stream is used for the virtual participant. If you enable individual participant recording, each physical participant is recorded separately. If you enable server-side composition, only the virtual participant appears in the composition.

Redundant ingest also enables continuous 24/7 streaming. IVS limits individual publishers to 24 hours, but with redundant ingest, IVS staggers the connection timeouts between the two physical streams and automatically switches which stream is used for the virtual participant, allowing subscribers to experience uninterrupted 24/7 streaming.

Requirements

Recommendations