When you hit Start Streaming, your video travels through a chain of protocols to reach viewers, and two names dominate that chain: RTMP and HLS. They do different jobs at different points, and confusing them leads to misunderstandings about latency, quality, and why your stream has a delay. Here is how each works and where it fits.
Two protocols, two jobs
RTMP (Real-Time Messaging Protocol) is a contribution protocol, it carries your feed from your encoder up to the streaming platform’s ingest server. HLS (HTTP Live Streaming) is a delivery protocol, it carries the stream from the platform out to thousands of viewers’ devices. In most modern setups you use RTMP to send and HLS is what viewers receive, often without either of you thinking about it.
Why the split exists
RTMP maintains a persistent connection ideal for a single uploader but scales poorly to millions of viewers. HLS chops the stream into small file segments served over ordinary web servers and CDNs, which scales almost infinitely and plays natively on browsers and phones. Each protocol is optimized for its half of the journey.
Head-to-head
| Aspect | RTMP | HLS |
|---|---|---|
| Role | Contribution (upload) | Delivery (playback) |
| Transport | Persistent TCP connection | Segmented files over HTTP |
| Latency | Low (seconds) | Higher (often 6–30s) |
| Scalability | Limited | Massive via CDN |
| Device support | Needs conversion to play | Native in browsers/phones |
Where your stream delay comes from
The multi-second gap between your action and what chat sees mostly comes from HLS. Segmenting video into chunks and buffering them for smooth playback inherently adds latency. Standard HLS can run 15–30 seconds behind; Low-Latency HLS and reduced buffer settings can cut that to a few seconds, which is why platforms offer low-latency modes.
What you actually control
As a streamer you configure the RTMP (or newer contribution) side, your server URL, stream key, and bitrate. The platform handles the HLS delivery. Enabling a platform’s low-latency option reduces the HLS buffering, tightening your chat interaction at a small cost to playback smoothness on weak connections.
The bigger picture
Newer contribution options like SRT are gaining ground for their resilience over unstable networks, and delivery is evolving with Low-Latency HLS and other formats. But the fundamental division, one protocol optimized for uploading, another for mass delivery, remains the backbone of live streaming.
Where SRT and low-latency options fit
RTMP is decades old and, while reliable on stable networks, it handles packet loss poorly. That is why SRT (Secure Reliable Transport) has grown popular for contribution, especially over shaky or long-distance connections, it recovers lost packets and tolerates jitter far better. If your upload path is unreliable, an SRT-capable workflow can noticeably improve stability compared with plain RTMP.
Reducing the delivery delay
On the HLS side, Low-Latency HLS shrinks segment sizes and buffering to bring the viewer delay down from the traditional 15–30 seconds toward just a few. Platforms expose this as a low-latency toggle. The trade-off is that very small buffers are less forgiving of a viewer’s weak connection, so ultra-low latency can mean more rebuffering for some of your audience.
What to actually configure
As a streamer, focus on the contribution side you control, your server or ingest choice, bitrate, and whether to use RTMP or SRT, and on the platform’s latency mode. Leave the HLS delivery mechanics to the platform. Understanding the split simply helps you interpret why delay exists and which knob to turn when chat interaction feels sluggish.
FAQ
Do I choose HLS or RTMP when I stream?
You generally configure the contribution protocol (RTMP or SRT). HLS is chosen by the platform for delivery to viewers, so it usually is not something you set directly.
Why is my stream delayed by 20 seconds?
That delay is mostly HLS segment buffering on the delivery side. Enable your platform’s low-latency mode to reduce it if tighter chat interaction matters to you.
Why is there a delay between my action and chat?
That delay comes mostly from HLS on the delivery side. Chopping the stream into segments and buffering them for smooth playback inherently adds latency, traditionally 15 to 30 seconds. It is not your RTMP upload causing it. If tighter chat interaction matters, enable your platform’s low-latency mode, which shrinks the HLS buffering to a few seconds, at the cost of slightly less smooth playback for viewers on weak connections.
Bottom line
RTMP carries your feed up to the platform; HLS delivers it out to viewers at scale, and HLS buffering is the main source of stream delay. Understanding the split explains latency and helps you use low-latency modes wisely.
