How Will Consumers Use Faster Internet Speeds?
freedom-to-tinker.com
freedom-to-tinker.com
The answer to this question, for me, is "Doin' the same shit. Slightly faster."
The speed test results are always fantastic, but real internet usage isn't quite as utopian as the hype makes it out to be. In some cases, I suspect there are bottlenecks between my house and whatever server I'm connecting to. In other cases, it almost feels like intentional throttling.
Netflix and Hulu performance is outstanding, but Amazon Prime videos occasionally drop into that super-blocky low bandwidth mode for seemingly no reason.
Most websites work great, but I've had super weird experiences with sites and servers that use IPV6 when I'm on my MacBook.
It's these weird quirks that probably keep me from noticeably changing my internet habits. But, as usual, YMMV.
With both OpenBSD and Debian, I've had no problems routing at gigabit speed locally and the maximum 350 Mbps on WAN.
Also, it looks like it would cost ~150->300 USD to make a complete (board, case, storage, AC adaptor) router?
What do you mean by firewall, exactly? I do nothing extreme, just an pf or iptables ruleset, run dnsmasq, privoxy - those sorts of things. Glad to take some performance benchmarks you can recommend.
The basic stuff: NAT translation, port forwarding, connection rejection and the like.
For bonus points, doing stuff like traffic shaping/prioritizing (like CoDel) at gigabit speed would be rather interesting.
At the moment the EdgeRouter Lite can do limited amounts of packet inspection using the offload hardware. As time goes on the Ubiquiti folks figure out how to better use the offload chip, but -for now- rate limiting and traffic shaping has to run through the thing's CPU, which -IIRC- gives you somewhere north of 100mbit/s of throughput.
> I think it's a good deal compared to something like...
Oh, for something that I would expect to be able to keep using for 10 years, I think it's quite a reasonable price. :D
Edit: Yeah, I don't really have any perf benchmarks to recommend. I guess -if I had the time and the gumption- I'd do something like set up iperf (probably 3 so you can use TCP) on two machines (each in a different subnet so packets would pass through the router) and adjust the iperf listen port so that my various firewall rules triggered, and compare performance.
I have a less than stellar connection for the US (~25/5-ish?), and getting my voip to work took some real tuning in PFsense. The move to PFsense was basically motivated by the fact that I couldn't even do the things I wanted to prioritize voip via the gui provided by the mainstream combo wifi/router I had before. Without tuning (firewall, qos settings), it would frequently sound terrible as soon as someone fired up something bandwidth intensive.
I wonder how much is HW vs. SW, since the router I'm using these days is a low-end single core (2x hyper thread) atom box from Soekris that rarely has >10% cpu load. There appear to be much faster HW options available from various vendors these days at a lower price point than I paid for the soekris when it was new, but PFsense wasn't available on my cheap MIPS wifi-router.
Anyone who wants to roll out a fast service should certainly take that cue.
And it was only recently that any routers started including CPUs that are powerful enough for high-speed packet processing. Prior to 802.11ac, basically everything was based off '90s-era single-core MIPS and the only hope of 100+Mbps throughput was to use hardware offloads that severely curtail the kinds of packet processing you can accomplish (and usually isn't supported by any open-source driver). Now we've got 1+GHz multi-core ARM processors in wireless router SoCs, but no mature drivers for them.
Here is the Edge Router Lite http://amzn.to/1UhB9SF
It's not a 'lazy developer' issue, but hardware really is the limiting factor. You would be surprised how much a router can impact your speed once you get above 50Mbps.
With Google Fiber "Free" and Time Warner before it, I was capped at 1 Mbps upstream and cloud services were much more or a chore.
A fast connection doesn't mean that I'm deciding to throw more stuff online or make more use of cloud services. It's just nicer that the stuff I have to transfer moves a helluva lot faster now. So I haven't noticed a significant difference in my usage.
My point is - 1gbit/s - not as utopian as it first sounds.
So any technical barriers to making all hdds obsolete are evaporating. I guess with the rise of ssds the only draw of hdds is now price, and I suppose that feature won't be supplanted by some service provider any time soon.
Fun story though, it was fast enough that I got kicked for an hour from my school's network because I was downloading way too much output data, way too quickly over SCP from the school supercomputer.
When we went from modems to "broadband" (>1mbps or so) it grabbed the good-ole-boy music industry by the throat and shook them so hard they've still got PTSD. >10 and the movie/tv industry had their reckoning. Consumers won huge both times. Can't wait to see whats next.
But currently the limits are more tied to physiological limits - the highest bandwidth channel for a human is the eyes and we know their limits (somewhere between 1080p to 4K depending on various things), and we've almost covered that.
The only open question is VR streaming, but i wonder whether this work considering the extreme low latency required ?
This goes counter to my experience and thinking. I am sure that this is true on a phone form factor but 8k TV is definitely better than 4k on a 60"+ screen (I've never seen them side by side on a smaller device and I am unsure how big the demo set up was).
Current 4k has limited colour capability, ramping up the stage to richer pallets will increase the bandwidth demand, as will high frame rates and the increase in resolution to 8k.
VR streaming is interesting but I guess that filming it will be a big challenge !
But if we only talk about resolution , 4K reaches the limits of the eye at around 1.1 meter [1], and most people sit farther than that from their tv.
But usually at display demos , they let you sit closer to the tv , this and other factors (compression, placebo, framerate , maybe brightness , etc etc ) might help create better experiences or illusions/salesmanship with 8K.
Another option is for rare few maybe the retina limit is higher .
[1]please use this calculatoe physiological limits of eyes r to see thhttp://isthisretina.com/
Quality has a quality all its own. Music piracy in the era of Old Napster (before they went legitimate) was terrible. Horrible mislabeled MP3s at a bitrate best described as "wax cylinder quality" with no album artwork or other extras buying a CD would provide. Plus, of course, you were stuck behind dial-up, and even poor-quality MP3s go at about a megabyte of data per minute of music, so you were looking at substantial download times over unreliable connections for anything approaching a full album.
It was wonderful only in that it worked at all, and that it had music record stores wouldn't touch, at least not if you were out in the boonies. It was one of the few pre-YouTube channels for obscure and out-of-print music. However, that didn't make it convenient in any absolute sense, and it was usually only high-quality compared to not having the music at all.
These days, with most torrent software, you can download a whole discography, in high-quality FLAC with full album images and so on, well-organized and correctly labeled, simply by copying-and-pasting a magnet link. Multiple megabyte downloads go by in seconds. The difference in the experiences is night and day.
The music industry got angry at Old Napster. They're being crushed by Bittorrent.
But sure, BitTorrent definitely had a bigger impact on the music industry.
Saw basically no difference in day to day usage. Even saw 0 difference in download speeds off of VPS servers etc.
Was hoping my employer's VPN would be like I'm on site (or as close as one could get). Bottle neck there as well.
When your average consumer is stuck on low speeds, imagine what the consequences are on those who rely on this for their work (not to mention for those who tinker and innovate in their "garage").
Where I live in Canada, even for your average business, it's either a barely sufficient DSL/cable line at 70-100$/month, or fiber which usually starts at 500$/month. It's ridiculously expensive to get good bandwidth. Sometimes it's easier to stick point to point antennas on the roof and connect directly to a datacenter.
Faster speeds: they will find a use for it, and everyone wins.
802.11ac can drop all the way back to 6Mbps under really bad conditions!
The big difference with faster speeds is that you don't have to muck about with QoS. Just throw bandwidth at the problem. This means you can actually sell a product that would depend on it, since support becomes manageable.
1 Mbps is adequate for VoIP. But at 10 Mbps I can use it without thinking. I can get an account with any provider, start simultaneous calls, and there are no dropouts even when torrenting.
100 Mbps is adequate for IPTV, in broadcast HD quality, when the streams are over a controlled network. But with 1Gbps it becomes just another service, even if my kids stream too.
The big difference, as someone else pointed out here, is when upload speeds catch up. Assymetric broadband is quickly ACK limited, which disturbs real time streaming in the other direction. Latency is key here, and just getting rid of that cable modem and hit the wire directly really helps.
Regarding piracy, I think it is obvious that it goes down when real time streaming gets practical. It's more convenient, really. So more broadband really does transform markets. The problem for the past ten years has been that it's not really deployed globally, and many of the rich western countries are stuck with cable modems because of lock in effects. This should be an obvious opportunity for an entrepreneurial spirit, just wire up a suitable area and you can be first to sell IP services to them. Cable TV and phone service are moving to IP, it's just a matter of time.
Don't forget the CSS "frameworks," gratuitous fonts, and gigantic auto-playing background videos, too.
It is fun to watch new customers bandwidth consumption. It maxes out for a few days to a week...and then then drops. As far as we can tell people back everything up, update everything, torrent everything .... and then run out of things to download / upload.
http://www.bufferbloat.net/projects/bloat/wiki/Introduction
'One of the most insidious things about bufferbloat is that it easily masquerades as something else: underprovisioning of the network. But buying fatter pipes doesn't fix the bufferbloat cascades, and buying larger buffers actually makes them worse!'
The fellow who wrote the article is one of the folks who performed the study.
See Facebook's auto-playing videos as an example.
I "self-throttle" data speeds by opting for for the lowest tolerable internet service to keep me under my cap.
F@#$ing Comcast.
ISPs have been (similar to yourself) using the limited access speeds to self-throttle the data usage of customers to limit their aggregated backhaul costs.
If I pay higher monthly connection costs for high speed access (the monthly service charge) and pay to actually use the service at the intended speed (additional fees generated by a data cap), I am being double billed for service.
Data caps in conjunction with a higher costs for high speed access are a predatory business model.
Thus, as Fiber rolls out to new areas, expect the pre-existing players to either improve their own services (to everyone's benefit) or try to undercut Google (to their own detriment).
That could even open the door to P2P multiplayer paradigms, which might make MMOs support more player in a same world.
Once you reach gigabit speeds, you can do all kinds of crazy cool things, like editing 4K videos remotely.
Even 10ms of latency can be a serious drag on your user interface. Thin clients kind of suck for this reason.
Because you actually own the physical machine; some may disagree, especially the DRM crowd, but there's a big difference between having hardware that is and on your property and just having volatile control over a machine in some unknown location.
Fun experiment: Try ssh-ing into a web host or a school account. Peruse a few directories and open a few files in VI. Type in a paragraph of text. Now try the same on your local machine's terminal and compare the responsiveness!