What adaptive streaming actually does

The short version

Adaptive streaming means your video is not one file. It is several, all the same film at different qualities, cut into small pieces. The player downloads a few seconds at a time and decides, piece by piece, which quality to take next.

That is the whole trick. Because the decision is made every few seconds rather than once at the start, the video can follow a connection up and down without ever stopping to rebuffer.

What that looks like in practice

Someone opens your link on fibre at home. The player takes a couple of seconds at a low quality to start instantly, sees there is plenty of headroom, and moves up to the top rung within a few seconds. They watch the whole thing in 4K and never notice the first two seconds were not.

Someone else opens the same link on a train. The player starts the same way, notices the download is slower than playback, and steps down. The picture softens. The video does not stop. When the train leaves the tunnel it steps back up.

Neither of them chose anything, and you did not upload anything twice.

Why not just send one good file

Because you do not know who is watching. Send everyone a 4K file and half your audience stares at a spinner. Send everyone a small file and the half on good connections get a soft picture on a large screen for no reason. A single file is a guess about a viewer you have never met.

What you have to do about it

Nothing, if your host does it for you. Upload the file once. We build the versions, cut them up, write the index the player reads, and serve the pieces. You get a link.

If you are building this yourself, the pieces are: an encoder that can produce several renditions from one source, a packager that cuts them into segments and writes an M3U8 or MPD index, storage, and a CDN in front. It is a weekend to get working and a long time to get right.