Skip to content

Latest commit

 

History

History
97 lines (57 loc) · 7.06 KB

README.md

File metadata and controls

97 lines (57 loc) · 7.06 KB
Raw

The server component - srtla_rec - in this repository is unsupported, no longer under development and not suitable for production deployment. Sign up for a BELABOX cloud account to benefit from the latest improvements, available on a global network of relay servers.

srtla - SRT transport proxy with link aggregation for connection bonding

This is srtla2, incompatible with previous versions of srtla. Remember to update srtla both on the receiver and the sender. srtla2 brings srtla_rec support for multiple simultaneous SRT streams, and many reliability improvements both for srtla_send and srtla_rec.

This tool can transport SRT traffic over multiple network links for capacity aggregation and redundancy. Traffic is balanced dynamically, depending on the network conditions. The intended application is bonding mobile modems for live streaming.

This application is experimental. Be prepared to troubleshoot it and experiment with various settings for your needs.

Assumptions and prerequisites

This tool assumes that data is streamed from a SRT sender in caller mode to a SRT receiver in listener mode. To get any benefit over using SRT directly, the sender should have 2 or more network links to the SRT listener (in the typical application, these would be internet-connected 4G modems). The sender needs to have source routing configured, as srtla uses bind() to map UDP sockets to a given connection. Only Linux is supported, however porting to other *nix platforms should be straightforward if needed. Update: srtla_rec has now been confirmed to work on Windows via WSL1.

Building (both on the sender and the receiver)

git clone https://github.com/BELABOX/srtla.git
cd srtla
make

This will produce 2 executables: srtla_send and srtla_rec.

Building the patched SRT (only needed on the receiver)

Our patched SRT library should be used on the receiver. srtla will work using the upstream SRT library, however performance may be degraded because the current upstream version causes excessive retransmissions when packets arrive out of order.

git clone https://github.com/BELABOX/srt.git
cd srt
./configure
make

See the SRT documentation for more information on building it.

Example usage

Let's assume that the receiver has IP address 10.0.0.1 and the sender has 2 (unreliable) modems with IP addresses 192.168.0.2 and 192.168.1.2 respectively, which can reach the receiver. We'll set up the srtla receiver to run on port 5000 and we'll use srt-live-transmit (from our SRT repository) in listener mode to make the stream available on 0.0.0.0:5001 for local or remote clients that use the upstream SRT library.

Receiver

path/to/srt/srt-live-transmit -st:yes "srt://127.0.0.1:5002?mode=listener&lossmaxttl=40&latency=2000" "srt://0.0.0.0:5001?mode=listener" &
path/to/srtla/srtla_rec 5000 127.0.0.1 5002

Notes: lossmaxttl is a required option to allow packets to arrive out-of-order without immediately sending NAKs to ask for retransmission. Its value is the size of the receive window. Values between 10 and 50 are probably a reasonable starting point. The NAKs sent by SRT are used by srtla to balance the traffic between the links and lower lossmaxttl values will create a stronger bias towards using the faster networks disproportionately. If the window is too small, that may cause excessive retransmissions and it may prevent link aggregation from working by sending most of the traffic through a single link. If the window is too large, it may prevent timely retransmission of lost / late / corrupted packets and therefore data loss. latency (in ms) will determine the time available for retransmission and packet reordering (together with lossmaxttl).

Sender

echo 192.168.0.2 > /tmp/srtla_ips
echo 192.168.1.2 >> /tmp/srtla_ips
path/to/srtla/srtla_send 6000 10.0.0.1 5000 /tmp/srtla_ips

With srtla_send running on the sender, SRT-enabled applications should stream to port 6000 on the sender and this data will be forwarded through srtla and srt-live-transmit to port 5001 on the receiver.

Note that instead of srt-live-transmit, you can directly use the end SRT application in listener mode on the receiver. It must be configured with the same options discussed above for srt-live-transmit and it should be linked against our modified SRT library.

Note that this basic setup doesn't implement authentication or encryption and srt-live-transmit can only handle one connection at a time.

Note that the sender should implement congestion control using adaptive bitrate based on the SRT SRTO_SNDDATA size or on the measured RTT. Also note that due to reordering, these values may be slightly higher during uncongested operation over srtla compared to direct SRT operation over one of the same network links.

How does it work?

The core idea is that srtla keeps track of the number of packets in flight (sent but unacknowledged) for each link, together with a dynamic window size that tracks the capacity of each link - similarly to TCP congestion control. These are used together to balance the traffic through each link proportionally to its capacity. However, note that no congestion control is applied.

This assumes that you're familiar with the SRT spec. TODO: technical description in more detail.

srtla v2

The main improvement in srtla v2 is that it supports multiple srtla senders connecting to a single srtla receiver by establishing connection groups - note that these are different from the experimental socket groups in SRT. To support this feature, I've introduced a 2-phase connection registration phase.

Normal registration:

  • Sender (conn 0): SRTLA_REG1(sender_id = SRTLA_ID_LEN bytes sender-generated random id)
  • Receiver: SRTLA_REG2(full_id = sender_id with the last SRTLA_ID_LEN/2 bytes replaced with receiver-generated values)
  • Sender (conn 0): SRTLA_REG2(full_id)
  • Receiver: SRTLA_REG3
  • [...]
  • Sender (conn n): SRTLA_REG2(full_id)
  • Receiver: SRTLA_REG3

Error responses are only sent from the receiver. If the sender encounters an error, it should just abandon the relevant connection group or connection, and it will be garbage collected on the receiver side after some time. Possible error responses are sent after receiving a SRTLA_REG1 or SRTLA_REG2 request.

  • SRTLA_REG_ERR - Can be sent in response to any SRTLA_REG command. Operation temporarily failed, back off and retry later.
  • SRTLA_REG_NAK - Sent in response to SRTLA_REG1. Operation refused, give up. Either incompatible or access denied. A human-readable error message may be appended after the header.
  • SRTLA_REG_NGP - Sent in response to SRTLA_REG2 with an invalid ID. Register the group again with SRTLA_REG1