[NNagain] Towards a speed of light internet
Dave Taht
dave.taht at gmail.com
Fri Mar 1 08:32:06 EST 2024
This paper was written in 2014. How far have we come?
https://arxiv.org/pdf/1505.03449.pdf
On Thu, Feb 29, 2024 at 11:02 PM rjmcmahon via Nnagain
<nnagain at lists.bufferbloat.net> wrote:
>
> Van Jacobson had a talk about content networking in 2012. His
> perspective is from a network view.
>
> https://youtu.be/Bc-4PJPxfVQ?feature=shared
>
> TV affiliates could have inserted their own ads but nationals (ABC, CBS
> and NBC) had the copyright to the content. It wasn't technology that
> stopped the affiliates but what came to be known as broadcast rights.
> And lots of case law. The biggest "affiliates" with access to content
> where in NYC, hence that's why they're there.
>
> A digression: It is interesting that Comcast is in Philly, though
> Comcast bought 30 Rock in NYC. The same Rockefeller that was the richest
> man to ever walk the planet. Ralph Roberts and Comcast origins came
> through Tulepo, Mississippi. Never bet against those who make things
> work for poorer communities. They most always win in the long run.
> Colleges that have an origin story of a YMCA sign (vs a $305M data
> center landmark building are ones not to bet against either. Boston
> really should lead the world in real broadband, which includes FiWi, by
> my judgment.)
>
> Today majors with mail, social networks, etc. have basically skirted
> copyrights in the transition from broadcast to unicast. And are doing
> the same with so-called generative AI. Of course, they and the end
> device mfgs are going to use encryption to control the content. Less
> case law required.
>
> The whole theory presented by Geoff of going back to a TV model is only
> true if we accept that we can't pay for digital things, rather it has to
> be paid from trillion dollar+ ad budgets. More Cheez Whiz so-to-speak
> but targeted. I like blue fake cheez so I'll get that.
>
> The idea that ten CDNs are going to dominate is a bit silly to me.
> Setting up servers is easier and easier. It's also why companies like
> Cloudflare are giving away QUIC stacks so they can be the CDN point that
> decrypts. I'm skeptical that the digital world is going to stop at
> HTTP3/QUIC. TCP seems good to me for many things, like new apps that
> don't need a CDN encrypt-er for ad delivery. Today's technology may not
> map to next week, particularly if it's all software.
>
> My take is that the innovation in sw & devices has been basically
> stalled since the mobile phone become the common person's computer and
> Steve Jobs passed. People are being trained on substandard devices and
> substandard mobile networks that, yes, are basically provided by the
> RBOCs who renamed themselves. The richest technologists tend to focus on
> data centers and don't want to make products for the common person. Just
> another epitaph per Thomas Gray and his Churchyard elegy.
>
> I think we are in a state of what's old is new again. Not all that
> innovative (though does make for trips to Thailand and talks about
> elephants I guess.) I don't think we have to repeat history all the
> time. Though behaviors like wars over religious beliefs does suggest the
> human condition struggles to overcome historical cultural values, which
> much of the time should be abandoned.
>
> We get to choose our epitaphs, not so much in words, but more so in our
> actions. I hope for our kids sake we each and all choose well. Pay a
> woman an honest pay for her work and don't ask advertisers to respect
> privacy and pay for things at the same time. Choosy mothers choose Jif
> no more.
>
> Bob
> > Get rid of the advertisers as the source of funds and everything
> > changes. This projection was obvious from how radio and television
> > rolled out in their days, being driven by Kraft TV and those soap
> > operas. The show was built for the ad, not the other way around. It's
> > also not a technology thing but a content creator's getting paid
> > thing, again all so an ad can be delivered. The dearth in high quality
> > content is that people will have to pay. We may never pay but that's
> > on us and our free rider thinking vs on technology or comm infra. We
> > said goodbye to journalists as Yahoo did in 2000 when asked if they
> > would ever pay one for the news articles they were "stealing" and the
> > young execs said, "No. I don't see why I would have to."
> >
> > Our communications networks have to move beyond entertainment and
> > social affirmations. They have to become life support capable. This
> > includes in home WiFi networks. Just as every power receptacle in a
> > house isn't for a TV, a network comm channel, port or whatever it's
> > called, is not only for a humans directly either, at least not per our
> > very limited 5 sensory system and our error prone brains.
> >
> > Machines and senors are going to be a major sea change but it's hard
> > work to get there, particularly when building off a last mile
> > infrastructure built for HBO and NFL.
> >
> > Or legacy isn't about advertising from the broadcast networks or about
> > some of the worst computers possible, something now called a smart
> > phone, which isn't true. It's not about AI, VR, AR etc either. It's
> > about communications as fundamental infrastructure and everything that
> > entails which goes way beyond ip addresses and CDNs. Fields like
> > linguistics come to mind.
> >
> > Predicting the future from the past is easy. Making a different future
> > other than the past is hard. What's old is new again until somebody
> > decides to actually innovate.
> >
> > Just my $0.02,
> > Bob
> >> On Thu, Feb 29, 2024 at 12:26 PM Lee via Nnagain
> >> <nnagain at lists.bufferbloat.net> wrote:
> >>>
> >>> On Thu, Feb 29, 2024 at 9:12 AM Dave Taht via Nnagain wrote:
> >>> >
> >>> > He is being incredibly provocative this week. It hurt to sit through this.
> >>> >
> >>> > https://www.youtube.com/watch?v=gxO73fH0VqM
> >>>
> >>> Yes, he's provocative - but also entertaining. And don't forget the
> >>> audience:
> >>
> >>
> >>> ABOUT APRICOT
> >>>
> >>> Representing Asia Pacific's largest international Internet
> >>> conference,
> >>> Asia Pacific Regional Internet Conference on Operational Technologies
> >>> (APRICOT) draws many of the world's best Internet engineers,
> >>> operators, researchers, service providers, users and policy
> >>> communities from over 50 countries to teach, present, and do their
> >>> own
> >>> human networking.
> >>>
> >>> His last slide deck seemed to be a call to arms. He's near the end
> >>> of
> >>> his career, so for all the Internet engineers, etc. I saw it as a
> >>> "here's where we're going. Do you want to contribute to this trend
> >>> or
> >>> take the Internet in a different direction?"
> >>
> >> I perceive the internet as a communications network, not just as a
> >> content one. Chat, email, and other bidirectional communications
> >> are the most useful parts of it, and cannot be cached.
> >>
> >>>
> >>> For example, after talking about CDNs and how most content is now
> >>> local he brings up the bit about if 10% of your traffic costs you 90%
> >>> of your carriage costs, if I was a rational provider, I would say to
> >>> all those customers who need that 10% of the traffic go find someone
> >>> else. I'm not going to do it. Don't forget, this is a deregulated
> >>> world - you can do that. There is no universal obligation to carry
> >>> default.
> >>
> >> So the audience was in the cheap seats in the back like me, were
> >> silently grinding their teeth?
> >>
> >>> Does network neutrality require an ISP to connect you to the Internet
> >>> at large? Or do they get to drop the "expensive" traffic that
> >>> requires connecting to a transit provider (or however they do it now
> >>> to connect to the global Internet).
> >>>
> >>> I was a bit dubious about the assertion that most traffic stays
> >>> within
> >>> the AS but surprise, surprise, surprise (most people here are old
> >>> enough to remember Gomer Pyle.. right?).. youtube content is in the
> >>> Verizon network. Start wireshark, get the IP address of the youtube
> >>> server and
> >>> $ sudo traceroute -6TAn 2600:803:f00::e
> >>> traceroute to 2600:803:f00::e (2600:803:f00::e), 30 hops max, 72 byte
> >>> packets
> >>> <.. snip ..>
> >>> 3 2600:4000:1:236::326 [AS701] 33.323 ms 2600:4000:1:236::324
> >>> [AS701] 2.542 ms 2600:4000:1:236::326 [AS701] 33.315 ms
> >>> 4 * * *
> >>> 5 2600:803:6af::6 [AS701] 3.843 ms 3.838 ms 3.834 ms
> >>> 6 2600:803:f00::e [AS701] 2.911 ms 2.216 ms 2.472 ms
> >>>
> >>> Do the same for Netflix and I get three [??] different ASs:
> >>> $ sudo traceroute -6TAn 2600:1f18:631e:2f84:4f7a:4092:e2e9:c617
> >>> traceroute to 2600:1f18:631e:2f84:4f7a:4092:e2e9:c617
> >>> (2600:1f18:631e:2f84:4f7a:4092:e2e9:c617), 30 hops max, 72 byte
> >>> packets
> >>> <.. snip ..>
> >>> 5 2600:803:9af::82 [AS701] 8.048 ms 2600:803:9af::5a [AS701]
> >>> 8.297
> >>> ms 2600:803:2::5a [AS701] 8.294 ms
> >>> 6 * 2620:107:4000:c5c0::f3fd:f [*] 2.846 ms
> >>> 2620:107:4000:c5c1::f3fd:20 [*] 2.810 ms
> >>> 7 2620:107:4000:cfff::f202:d5b1 [*] 8.148 ms
> >>> 2620:107:4000:cfff::f203:54b1 [*] 5.289 ms
> >>> 2620:107:4000:cfff::f202:d4b1 [*] 4.300 ms
> >>> 8 2620:107:4000:a793::f000:3863 [*] 4.865 ms
> >>> 2620:107:4000:a610::f000:2403 [*] 5.245 ms
> >>> 2620:107:4000:acd3::f000:e060 [*] 5.201 ms
> >>> 9 * * *
> >>> 10 2600:1f18:631e:2f84:4f7a:4092:e2e9:c617 [AS14618/AS16509] 4.881
> >>> ms 4.864 ms 4.848 ms
> >>> 11 2600:1f18:631e:2f84:4f7a:4092:e2e9:c617 [AS14618/AS16509] 6.351
> >>> ms 6.075 ms 5.935 ms
> >>>
> >>> Does it violate network neutrality that youtube content takes the
> >>> "fast lane" getting to me?
> >>>
> >>> and just for chuckles..
> >>> $ dig 2024.apricot.net aaaa +short
> >>> 2001:dd8:f::1
> >>
> >> Anycast technology can certainly be applied to more parts of the
> >> internet than it is today.
> >>
> >> QUIC tho, seems to enable the idea that all of google could run off of
> >> 8.8.8.9, all of cloudflare, 1.1.1.9, etc.
> >>
> >>>
> >>> $ sudo traceroute -6TAn 2001:dd8:f::1
> >>> traceroute to 2001:dd8:f::1 (2001:dd8:f::1), 30 hops max, 72 byte
> >>> packets
> >>> <.. snip ..>
> >>> 3 2600:4000:1:236::324 [AS701] 27.390 ms 2600:4000:1:236::326
> >>> [AS701] 5.711 ms 2600:4000:1:236::324 [AS701] 27.384 ms
> >>> 4 * * *
> >>> 5 * * 2001:2035:0:bb3::1 [AS1299] 7.235 ms
> >>> 6 2001:2034:1:73::1 [AS1299] 7.763 ms 6.033 ms 5.996 ms
> >>> 7 2001:2034:1:b7::1 [AS1299] 11.530 ms 2001:2034:1:b8::1 [AS1299]
> >>> 10.704 ms *
> >>> 8 * * *
> >>> 9 2001:2000:3080:230d::2 [AS1299] 72.609 ms 72.594 ms 73.096 ms
> >>> 10 * * *
> >>> 11 * * *
> >>> 12 * * *
> >>> 13 * 2402:7800:10::2 [AS4826] 289.033 ms *
> >>> 14 2402:7800:10:1::12 [AS4826] 290.608 ms 292.440 ms 290.840 ms
> >>> 15 2402:7800:10:8::16 [AS4826] 228.836 ms 229.406 ms 231.379 ms
> >>> 16 2001:dd8:8:38::2 [AS4608] 233.803 ms 231.332 ms 233.572 ms
> >>> 17 2001:dd8:f::1 [AS4608] 231.822 ms 231.137 ms 232.772 ms
> >>>
> >>> Oh my.. I'm betting that's a lot more than 100 miles away :)
> >>>
> >>> Regards,
> >>> Lee
> >>> _______________________________________________
> >>> Nnagain mailing list
> >>> Nnagain at lists.bufferbloat.net
> >>> https://lists.bufferbloat.net/listinfo/nnagain
> > _______________________________________________
> > Nnagain mailing list
> > Nnagain at lists.bufferbloat.net
> > https://lists.bufferbloat.net/listinfo/nnagain
> _______________________________________________
> Nnagain mailing list
> Nnagain at lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/nnagain
--
https://blog.cerowrt.org/post/2024_predictions/
Dave Täht CSO, LibreQos
More information about the Nnagain
mailing list