Slashdot Mirror
India's Telecom Regulator Backs Net Neutrality (reuters.com)
From a report on Reuters: India's telecom regulator has made recommendations to ensure an open internet in the country and prevent any discrimination in internet access in a long-awaited report (PDF), after debating the issue of net neutrality for more than a year. The Telecom Regulatory Authority of India (TRAI) said it was not in favor of any "discriminatory treatment" with data, including blocking, slowing or offering preferential speeds or treatment to any content. The Indian regulator's support of net neutrality stands in contrast to the recent stance taken by the chair of the U.S. Federal Communications Commission. Last week, Ajit Pai, a Republican appointed by President Donald Trump in January, unveiled plans to rescind so-called net neutrality rules championed by former President Barack Obama that treated internet service providers like public utilities.
Most of us have at least a vague understanding of why some companies wanted NN destroyed, and how it will negatively affect American Internet services from the perspective of the average user.
How many liars will come out of the woodwork this time to support the NN repeal by sharing 'alternate facts' and tell us history didn't happen the way it's documented?
India gets it right where an Indian origin American doesn't.
> Net Neutrality, is an easy set of rules to follow. Treat every packet like any other one.
Treating low-bandwidth flows that extremely sensitive to jitter (ex VoIP) the same as you treat high bandwidth flows that don't care about jitter (ex Netflix) would make everything work poorly. If all packets were treated the same, 1960s-style, your video would stutter so bad it would be unwatchable in 352 Ã-- 288 and your VoIP would have sounds arriving in the wrong order, so "hello jellomizer" would arrive as "lloeh mijellozer". Fortunately, the hundreds of pages of NN rules that were in effect for a year and half weren't quite that stupid.
An early draft DID say spam and spoofed packets had to be treated equally as legitimate packets, but we got the most obvious cases of stupid fixed. For security and prevention of DOS it's important to block spoofed and malicious packets as close to the source as possible and that wasn't completely made illegal by the NN rules as approved in 2015. It was damaged, but not completely barred in all cases like "treat every packet the same" would do.
People who don't know a port from a fragment do say "treat everything exactly the same" because they don't have any idea how carrier networks work, but policy makers at the FCC aren't listening to Facebook meme-based comments. One Congresswoman did, dangerously, but that was handled.
Unfortunately, there are several books each a thousand pages or more teaching network optimization methods to get the right flows into the right paths in the right order to provide the desired combination of jitter, packet loss, latency, and bandwidth that's needed for each type of flow. We moved away from mostly packet-based policy a long time ago - each packet is part of a flow. The user experience is based on the flow, such as a video stream or http response, so we focus most of our attention on flows, rather than tiny artificial, arbitrary pieces of data within the flow (packets). In other words, making routing, policing, and queueing decisions a million times, for a the million separate packets in a video, would be stupid - instead we make those decisions once, based on "he wants to watch a pre-recorded video, which will be buffered". Packet level stuff mostly comes into play because we know real-time flows such as RTP tend to have much packets that need to arrive in order, at equal intervals, while YouTube videos have much larger packets that are fine with being delivered in bursts, because it's buffered anyway.
Anyway the point is, I only know maybe 15% of what there is to know about how to most effectively provide the best network performance for each thing you want to do on the internet, because I've only read maybe 4,000 pages on the subject. The best engineers, the CCIEs, apply tens of thousands of pages worth of knowledge to get the right packets where they need to be at the right moment, with the right volume of packets vs the acceptable packet loss.
It's a complex science. Rules telling them how they must do it must take into account the complexity. Not only the complexity of the network technologies that are common in carrier networks today, but emerging technologies. Otherwise, the rules will be either damaging, ineffective, or both. (And the 2015 included plenty of both)