Several layers cooperate, and each answers a different question. IP addressing decides where packets go. On carrier-managed networks, multicast with IGMP lets one live feed serve many households at once, which is efficient but requires the operator to control the whole path. Over the open internet that control does not exist, so services use unicast HTTP segment streaming: the file is split into short pieces listed in a manifest, and your player requests them one after another over ordinary web connections. RTSP still appears in on-demand and camera systems. The practical upshot is that internet protocol television over the open internet behaves like a very disciplined web download, which is why any device with a browser-grade network stack can receive it.

Your player keeps a queue of downloaded segments and plays from the front while filling the back. With roughly six-second live segments and a shallow queue, the tolerance for a late delivery is a handful of seconds. Anything slower than that shows on screen as a stall. This has a diagnostic consequence most explanations skip: if you switch away from a channel and back, and it plays normally, the session stalled rather than your bandwidth running out, because a genuinely saturated line would fail again immediately. On-demand playback hides the same faults because the player is allowed to run minutes ahead, so brief interruptions never reach the visible edge of the buffer.

A 1080p stream runs about 5-8 Mbps and uses roughly 3 GB per hour. A 4K stream runs about 15-25 Mbps and roughly 7 GB per hour. Codec choice moves those numbers as much as resolution does, since HEVC delivers a comparable picture at about half the bitrate of H.264. Two figures matter when sizing a line. First, simultaneous connections multiply the requirement, so a household running two 4K streams wants 30-50 Mbps of headroom. Second, monthly data caps are hit by hours watched rather than by speed. Ten hours a week of 4K viewing is roughly 280 GB a month, which is worth checking against your plan before you blame a service for stuttering.

Carrier-managed internet protocol television reserves capacity on its own path, so congestion is controlled and picture quality is predictable. The costs are that you must buy the carrier's line, accept its box, and stay inside its footprint. Delivery over the open internet gives up that reserved path and gains portability: any connection in 190+ countries works, the device is yours, installs are unlimited and plans cover 1-5 simultaneous connections. The honest tradeoff is that quality now depends on two things you should measure separately, your own line and the provider's capacity. That is precisely why a stated uptime figure matters, and why ours is published as 99.99%, or about 53 minutes a year.

Ask for numbers and then check them against your own eyes. Request the uptime figure; a service that answers with adjectives instead of a percentage has answered the question. Test on a live event, because a shallow live buffer exposes capacity problems that on-demand playback conceals. Watch the payment path as well as the picture: crypto-only checkout, refunds issued as gift cards or store credit, and a refusal to sell any trial are the signals people most often name after losing money in community threads. A $5 trial that runs 24 hours, a 7-day money-back window on plans, no auto-renewal and no stored card are all checkable before you spend more.

1. 1Get the uptime figure as a percentage
2. 2Test during live sport, not during a film
3. 3Check how refunds are paid before you pay anything