Could Netflix change its video streaming service to use a P2P architecture, in order to save money on content delivery and sidestep peering conflicts with ISPs like Comcast?
That’s a possibility raised by Netflix CEO Reed Hastings in a blog post Thursday, which urged the FCC to make peering part of new net neutrality regulations. ISPs want Netflix to pay for delivering traffic to their customers because the company doesn’t consume as much traffic as it delivers — to which Hastings replied:
“Interestingly, there is one special case where no-fee interconnection is embraced by the big ISPs — when they are connecting among themselves. They argue this is because roughly the same amount of data comes and goes between their networks. But when we ask them if we too would qualify for no-fee interconnect if we changed our service to upload as much data as we download (in other words, moving to peer-to-peer content delivery) — thus filling their upstream networks and nearly doubling our total traffic — there is an uncomfortable silence.”
This brings up an interesting question: Could Netflix actually do that?
Could a service like Netflix stream videos via P2P?
P2P is best known for file sharing — think Napster’s MP3 swapping and movie downloads from the Pirate Bay, or even licensed torrent downloads, courtesy of BitTorrent Inc. At its core, it just means that users don’t access data from a central server, but instead exchange it between one another — and that same technology can easily be used for video streaming as well.
In fact, Chinese video services used P2P as their primary distribution mechanism for video streams for years. The Chinese internet was traditionally fragmented, with infrastructure being centered around a few major state-owned telecommunications companies. Reaching consumers with adequate speeds to stream video would have required significant investment from video service providers, which is why many of them decided to distribute P2P streaming clients instead.
Services like PPStream, PPLive and Xunlei all used their own P2P software, and even major broadcasters like CCTV used P2P to reach millions of viewers during major sporting events with higher reliability and lower costs than a server-based architecture could have afforded them. Only in recent years has there been a trend toward central architectures for some of these offerings.
In the U.S., P2P was also used for some time to power video streaming for CNN and others, but falling bandwidth costs and the unwillingness of consumers to install plugins or clients for streaming led most services to switch to a central architecture. Most recently, BitTorrent shut down its efforts to bring P2P live streaming to desktop PCs, and decided to focus on mobile devices instead.
Would P2P really double Netflix’s traffic?
Hastings suggested Thursday that P2P would “nearly double” Netflix’s traffic. That assessment was obviously meant to put pressure on ISPs, and a closer look shows that the math isn’t all that clear.
When a Netflix subscriber watches an episode of House of Cards in HD, he consumes about 3GB of data. If that same subscriber were also to upload that very same data to someone else to distribute it in a P2P fashion, it would lead to a total consumption of 6GB. Right?
Well, not so fast. First of all, by getting the data from the first user, the second subscriber wouldn’t access House of Cards from Netflix’s servers, which would mean that in total, about the same amount of data would change hands. And in reality, there wouldn’t just be two people watching the same content, but likely thousands, ideally leading to only incremental data consumption increases for each consumer. With a slightly larger overhead, there would be some traffic increase, but it’s very unlikely that this number would approach 100 percent.
Peering and the last mile: So close, yet so far
The real question here isn’t whether the total amount of bits caused by Netflix viewing would increase, but what the impact on peering as well as the last mile would be. Hastings suggested that switching to P2P could essentially lead to a world in which Netflix viewers would send as much traffic from an ISP’s network to other networks as they would consume. The real impact on peering would largely depend on the P2P architecture used.
Back when BitTorrent and other file-sharing technologies had a larger impact on ISP networks, some P2P developers banded together to propose a technical solution for this very problem. Dubbed P4P, it gave ISPs a way to steer the flow of file sharing traffic to make sure that users connected to geographically closer peers, or peers on networks that allowed them settlement-free peering. So if Netflix and ISPs cooperated, they could make P2P work — but given the current situation, that’s a big if.
The other pain point is the last mile. Back in 2008, Comcast admitted to throttling BitTorrent. It argued that file sharers were consuming too much bandwidth on the local level, causing network congestion for their neighbors. Comcast eventually moved away from these measures and towards data caps, and BitTorrent changed the protocol of its clients to be more aware of the state of the network and yield to other traffic. But if Netflix flipped the switch on P2P tomorrow, it could put lots of stress on the last mile, which could be the real choke points for ISPs.
What about mobile and TVs?
One of the challenges for Netflix would be that more than 80 percent of its traffic comes from mobile and connected devices. Distributing a P2P plugin to PCs is relatively simple, but making it work on the Xbox One could be significantly more challenging. P2P has been done on mobile devices, and adding a P2P component to Netflix’s mobile apps should be possible, even though issues like data caps on mobile plans as well as battery life would have to be addressed.
But the real issue could be making this work in the living room, where Wi-Fi could become another choke point. Consumers frequently use older networking equipment they got from their ISPs, and getting adequate bandwidth for HD video streaming is already a challenge for many. Now imagine that their smart TVs were also uploading bits and pieces as they streamed Orange is the New Black, and you can see that they’d frequently end up with congestion in the home. Some consumers might go out and finally buy a new router, but many would just blame Netflix if their streaming looked worse.
In the end, Netflix switching to P2P is nothing more than an academic exercise. Yes, it would be possible, and yes, it would save the company some money. But with the large number of Netflix users and the wide variety of devices they use to watch Netflix, P2P would also bring up a whole range of new problems.