> and says that ISPs may not attempt to evade net neutrality protections by slowing down traffic at network interconnection points
Here's an anecdote. Over 10 years ago, I worked for an ISP company in Russia. That was before they went draconian on the 'Net (beyond SORM-2, I mean) - I left around the time Russian government started to mandate blocking websites. Just to clarify - that was an age of rapid growth and users had a choice of ISPs (I had 7 possible to pick from), so providers had competed really seriously.
Which is why net neutrality wasn't a thing - on the contrary, providers were incentivized to set up local FTP servers. Uh, y'know, for warez, though, of course, officially ISPs just "didn't knew" what users were doing there - but that was before streaming services; even Steam was quite a novelty at the time. Those FTPs had faster access speeds (traffic shaping was turned off for their IPs entirely, so up to 100Mbps when 10Mbps connectivity was the norm) and unaccounted traffic. It was mutually beneficial - providers had saved (a lot) on uplink utilization and customers had exchanged content at much faster speeds. In '00s there were even citywide LANs communities, which were torn down only because of malware onset. Those were wild times.
Doing some funky shit like throttling certain traffic meant your users would be gone (people tolerated for a short while, then voted with their wallets). But we had to do this twice. First time it was disaster management, when the first uTP implementation happened. Users had updated their uTorrents and it was buggy and caused very high packet rate (and this is UDP we're talking about), so our edge routers started to choke on this UDP BitTorrent traffic. We had to identify those packets and push a rule that dropped them entirely. Second time was also about BitTorrent - in an university dorm, where we contracted to set up WiFi coverage - I don't remember the details but we had really struggled ensuring fair bandwidth distribution, as someone torrenting made network barely usable for others (because of radio interference, not bandwidth limitations), but we had to tune it down to make it work somehow.
Another story from those times. We had started to receive tons of complaints that YouTube was performing poorly. We've spent days trying to figure out what the heck was wrong there, and realized the problem was only manifested through one of the upstream providers. They had no clue either, so the temporary solution was to push a a bunch of excessive AS-PATH elements in BGP announce to deprioritize routing YouTube traffic from that direction, then wait until someone upstream figures it all out.
And yet another story is that we ran analytics (anonymized, we only looked at very coarse AS level traffic distribution) to understand where we talked to most and if we should peer with someone else to improve connectivity in those directions.
And, of course, we had blocked ports, such as tcp/25 or tcp/445. Would've been a disaster if we wouldn't have.
The morale of this is that as an ISP sometimes you have to violate neutrality in the interest of your customers, to ensure quality of service. Well, to think of it - QoS is inherently non-neutral since the whole point of it is to classify traffic and differentiate it. And sometimes it's hard to draw a line on what's fair and what's not. I mean, I questioned some things we did and only justified them as "well, I don't see any other options here to make it work". And I don't condone nor condemn those examples - just present them as an anecdote to say that there are always some nuances and sometimes things aren't exactly black and white.
Though when people speak about NN in the US they talk about shitty anti-customer moves like throttling Netflix trying to squeeze some money out of this.