How does the provider determine video viewing on YouTube?

Question

Computer networks

Many providers, for example, mobile ones, now provide such a service as "unlimited video viewing on youtube.com". Calling and clarifying the details, I was informed that the traffic will not be spent on watching the video . Those. if you just surf the site, then the traffic will be spent. Here , for example, is a typical service from tele2. And in the other conditions tab they write, I will quote:

As part of the service, only video traffic is provided unlimited, while updating the application, navigating the site and the application will consume the Internet traffic package included in the tariff.

I ask for what this service is, when I view it in a browser (and they also allow it in a browser), no matter how I tried to configure my browser, the traffic was still spent. And in connection with this question: how exactly can the provider determine that I am watching streaming video, because for streaming video, even though it happens through JS, the browser accesses the youtube.com domain ? It seems that YouTube has no specific domain for the stream. And the connection itself is organized via https.
Or does YouTube have a dedicated section for video streaming? Who knows, share at least superficially)
Well, can you give a couple of tips on how to adjust the browser or what to check when accessing YouTube?

Thanks

Answer 1

This is achieved using DPI (to detect and suppress unwanted traffic, for example, voice termination via Viber, torrents, etc.). As for chradjing (traffic payment), it's elementary by hostname / IP. Youtube, Facebook and others have an agreement with the operator - they provide all their resource identifiers, the traffic from which the operator evaluates to 0.
I comment in response, since
gogalikob4 has gone so far, do not put it in a comment. I say again, traffic is classified on the operator's GGSN / DPI (by IP addresses). GGSN/DPI uses the Gx protocol to PCRF to find out which Rating Group to route this traffic for a particular subscriber/client. If the PCRF has a profile entry for a specific subscriber (the "Unlimited Facebook" package, for example), then it sends a PCC-Rule Install response to Gx. GGSN/DPI through the mapping Rule->Rating Group determines the desired RG and uses it in the Gy protocol, which links GGSN/DPI and OCS (Online Charging System), if the payment control implies an online mode (alternatively, a CDR is written that has these data, and which goes to offline billing). OCS (or offline billing) for this RG has a set price of 0.
This is what happens at the most superficial level. But if you are interested in all this, then you should already be interested in and read special documentation (Gx/Gy/Sy/Gn/GTP protocols, traffic rating principles).
As for what is written:

spoiler

Расходование пакета интернета происходит, если на сайте или в приложении «Youtube» открываются страницы, содержащие ссылки и для отображения которых требуется соединение с внешними ресурсами (другими сайтами), либо для доступа в интернет используется браузер, сжимающий трафик, или прокси-сервер, либо при просмотре используют компрессию данных, анонимайзеры и VPN

It is clear that traffic is free if it comes from Youtube IP. If you somehow open content from another resource in the application, then it will not be from this IP - that's why such traffic will be paid. Same thing about VPN - the VPN server will receive from Youtube, and your device from the VPN server whose IP is not Youtube's IP. For other proxies. compressors, anonymizers, the same - you will not receive content from Youtube IP, and therefore it will be paid.

Answer 2

Perhaps we are talking about this
https://ru.wikipedia.org/wiki/Google_Global_Cache

Answer 3

There is a suspicion that for uploading videos from YouTube, there are simply separate ranges of IP addresses that are no longer used for anything. The provider knows about them and traffic from them does not charge.