I’ve been working with SRT links for years in studios and headend environments, and recently I came across something quite interesting that I thought might be useful to share here.

When you feed an unstable SRT input (bitrate spikes, packet loss, jitter, etc.) into a gateway and repeat it as UDP/RTP multicast without any remuxing at all, the multicast output stays surprisingly stable — even when the SRT input is suffering short periods of 8–12% packet loss.

As long as SRT manages to recover the missing packets within the buffer delay, the UDP/RTP output remains clean and unaffected.
No added latency, no TS modifications, no glitches.

I tested this across multiple setups (including a multi-NIC on-prem system we use for some cable headends) and got the same result every time:
The multicast side receives a perfectly solid stream even while the input SRT looks like it’s “fighting for survival”.

So I’m curious:

Has anyone else here noticed this kind of behavior?

Do you rely on SRT recovery to protect your multicast distribution layer?

Do you use dedicated SRT→UDP gateways, or feed SRT directly into your decoders?

Interested to hear real-world experiences from people running SRT contribution/distribution in production.