Weekly Edition
Return to the Press page
| |||||||||||||
What could you do with fat fiber? (Linux Journal)
Doc Searls wonders about
fiber. "Two years ago, Bob Frankston wrote Why Settle for Just
1%? while in the midst of his ramp-up as a Verizon FiOS customer. The
question is still on the table. I'd like us to help answer it by re-phrasing
the question: What could we, as Linux developers and users, do with fiber to
our homes and businesses?"
(Log in to post comments)
What could you do with fat fiber? (Linux Journal) Posted Aug 4, 2007 6:28 UTC (Sat) by tavis (guest, #14187) [Link] How about transmitting holographic images for 3D conferencing? But then someone will have to develop a cheap display, and besides someone else has probably already patented whatever algorithms are necessary to compress the data appropriately.
People are missing the point Posted Aug 4, 2007 9:21 UTC (Sat) by khim (subscriber, #9252) [Link] Fiber is not copper! Today it's impossible to offer more then 1% of fiber. Not hard, but impossible. 10Mbit ? Easy. 100Mbit ? No problem. 1Gbit ? As you wish. 10Gbit ? Hard but doable. 100Gbit ? Very expensive (need some expensive experimental hardware) but it's still the same fiber and it's still less then 1% of said fiber! Fiber can deliver 100-200Tbit easily! Why Verizon is not offering that ? Easy: the limitation is not fiber. Limitation is copper and silicon. We still don't have optical computer so every single byte which passes over fiber network should be handled by a lot of copper/silicon devices: routers, switches, etc. And this is where limitation is today. And copper/silicon will limit fiber for many-many years. So if you take in account reality of the situation then you'll see that switch from 10Mbit to 1500Mbit will not change the bandwidth available for other ventures (TV, phone, etc). What will it change ? Why, the price of routers, switches and other hardware. Now the question becomes: "why Verizon is installing routers and switches only good for 15Mbit network, not for 15Gbit network?" and not "why Verizon keeps 99% of bandwidth to itself?". And the answer to THIS question should be obvious: Verizon is commercial entity, it needs profit. Of course this answer raises the next question: if the fiber does not care about amount of data it carries (at least it'll be true for many years) and the real problem lies in "highway" it connects to... why can not you use fiber of one provider to connect to another provider (like with phone lines). But this is totally different question...
People are missing the point Posted Aug 4, 2007 15:24 UTC (Sat) by ajross (subscriber, #4563) [Link] Uh... Verizon is offering 15Mbps right now. That is 1% of 1.5Gbps, which by your classification scheme in the "As you wish" category.
Regardless, simply offering a 54Mbps link would be as much as could reasonably be used right now by common hardware.
People are missing the point Posted Aug 4, 2007 23:47 UTC (Sat) by khim (subscriber, #9252) [Link] Uh... Verizon is offering 15Mbps right now. That is 1% of 1.5Gbps, which by your classification scheme in the "As you wish" category. And what does it prove ? That actually Verizon has the capacity to offer you 1.5Gbps ? Of course it has! It has the capacity to offer you 15Gbps or even 150Gps. The question is simple: who'll pay for it ? If you have network distributed over wide area then your packet will pass 10-15 routers while traveling from one point to another. Each can handle no more then 200 such high-speed users (if even that: 720Gbit backplane, 40% utilization) and we need 10 of them... so ~20 users per router. That's $1000 per user! And that's just backbone routers... there are a lot of other hardware needed to process this traffic. So it's easy to see that there are few choice: I somehow fail to see ISPs choosing option 3 :). If/when the price of routers (and other hardware) will go down - faster options will become available, but as you can see problem with fiber is not fiber, it's copper/silicone in hardware required to handle all this TCP/IP traffic... Regardless, simply offering a 54Mbps link would be as much as could reasonably be used right now by common hardware. We'll see such speeds offered in next few years. By then 100% will be 5.4Gbit, so it again will be "just 1%" :-)
People are missing the point Posted Aug 7, 2007 11:17 UTC (Tue) by ekj (guest, #1524) [Link] People don't commonly use a large fraction of the bandwith available to them around the clock. Atleast residential users do not.
I have a 10Mb symetrical internet-connection, which is actually the lowest speed offered by my ISP (the alternatives are 25 or 100Mbps), so in theory I could download and upload aproximately 100GB/day, or 3TB/month if you prefer.
Actual usage is more typically 100MB/day though, even after you factor in the occasional days with a new Linux-distro slurped or similar, overall utilization is certainly no higher than 1%.
People are missing the point Posted Aug 7, 2007 11:33 UTC (Tue) by khim (subscriber, #9252) [Link] People don't commonly use a large fraction of the bandwith available to them around the clock. Atleast residential users do not. May be not in your area. may be not now. But what will you do when they'll start to actually use offered bandwidth ? If you infrastructure is not up to the task then you'll see slowdowns and other problems. If 100Mbit bandwidth will be available people will switch from 700Mb XviDs to 25GB Blue Ray rips. Or they'll start using HDTV videoconferencing. Or they can start using Google as local storage. Or any other such usage. You can not build you strategy around the idea that residential users will not actually use offered bandwidth. But all such discussion gone waay to far from the article. In regard to it you answer is good as mine: you say it's not a big deal if someone will offer high-speed channel - that people will only use 1-2% of offered bandwitdh anyway. I do not think it's true, but even if it's true it still answers the question raised in article: We MUST "Settle for Just 1%" - forcibly or voluntarily. Alternatives are unworkable...
People are missing the point Posted Aug 7, 2007 13:05 UTC (Tue) by ekj (guest, #1524) [Link] You can not build you strategy around the idea that residential users will not actually use offered bandwidth. Actually, you *must*. I didn't say I don't use it. I said I don't use it 24x365. That is the key. My *average* bandwith-consumption is only 1% of my max, this does however NOT mean that I'd be equally well-served with a link with a capacity only 1% of my current.(for me, upgrading from 1Mbps to 10Mbps was a major improvement, despite me using less than 100Kbps on *average*.) Any ISP not underprovisioning would go instantly bust. There's nothing special about networking. Everything, pretty much, works that way: Everyone with a car can drive anytime they please. Nevertheless if *everyone* with a car started driving it 24 hours a day, 365 days a year, the road-system would completely collapse, it has capacity for only a modest fraction of the existing cars. Everyone can take a shower anytime they choose. Nevertheless, if everyone started letting all their taps run 24x365, our water-supply would not hold up, not even close. Provisioning as if everyone will use 100% of what they *can* use, 100% of the time is insane, and not just when networking is the topic.
People are missing the point Posted Aug 7, 2007 17:37 UTC (Tue) by khim (subscriber, #9252) [Link] Actually, you *must*. I didn't say I don't use it. I said I don't use it 24x365. That is the key. If you want to use 100Mbit sometimes but 1Mbit on average your plan should say so OR it can include price for traffic (it's mostly the same thing). If you offer "unlimited 100Mbit" you should be ready to offer "unlimited 100Mbit". Everyone with a car can drive anytime they please. Nevertheless if *everyone* with a car started driving it 24 hours a day, 365 days a year, the road-system would completely collapse, it has capacity for only a modest fraction of the existing cars. Nope. It'll not collapse at all: gas price will skyrocket and that will be it. Everyone can take a shower anytime they choose. Nevertheless, if everyone started letting all their taps run 24x365, our water-supply would not hold up, not even close. And that's what countries with flat rate for water-supply found the hard way: there are a lot of waste and shortage problem all around the world. Provisioning as if everyone will use 100% of what they *can* use, 100% of the time is insane, and not just when networking is the topic. Think that people will use anything "in sane amounts" if not pressed in insane. This is proved again and again - and not just in networking. The shorter the feedback loop, the faster the response.
What could you do with fat fiber? (Linux Journal) Posted Aug 4, 2007 9:58 UTC (Sat) by jengelh (subscriber, #33263) [Link] >What could we, as Linux developers and users, do with fiber to our homes and businesses?
Download linux kernel tarballs with some decent speed. Currently, it still takes 125+ seconds for the regular German home user. ;-)
What could you do with fat fiber? (Linux Journal) Posted Aug 4, 2007 11:28 UTC (Sat) by pjdc (subscriber, #6906) [Link] Then updating your git tree should be very fast indeed.
Lots to do with fat fibre Posted Aug 4, 2007 11:54 UTC (Sat) by gdt (subscriber, #6284) [Link] Once you have 'enough' bandwidth, say 1Gbps, another set of factors comes into play to restrain network performance. Simple latency is one. Unneeded round-trip times in applications are disliked by network engineers just as applications which make unneeded system calls are disliked by the kernel folk. TCP itself performs poorly once the bandwidth-delay product gets near cross-country gigabit. So one thing that people can do with cheap large bandwidth is to help each other reduce latency: content distribution networks, P2P, and other overlay networks. Having such cheap large bandwidth also makes sharing the results of computation more economic. Say you're down-rating a HDTV broadcast to SDTV. You can share the results of that intensive computational load to other viewers of the broadcast that might desire a SDTV feed. Of course, it's a short step from there to full distributed computation, as with the Grid. Although that brings significant trust issues in some applications -- your neighbour may give malicious results. Cheap large bandwidth also moves the line between what is "inside" the computer and what is "outside". Is large mass storage needed "inside" the computer anymore? At the least there's a lot to be said for using a RAID mirror with the one disk located across the network. At 10Gbps the network finally rivals sending a station wagon full of tapes for throughput. So there's no reason not to do all archiving across the network. Large bandwidths also allow distributed sensor networks. The Square Kilometre Array is a massive and early example, but more mundane applications are obvious. If applied to surveillance cameras and microphones these networks would provide a serious intrusion into privacy. Large bandwidths also allow existing applications to be larger. As the Access Grid and video instant messaging shows, teleconferencing can grow from one-on-one to many-to-many. Quality can also increase, this can be marginal or dramatic. For example, the improvement in experience moving teleconferencing from SDTV to HDTV is remarkable -- someone can hold up a printed document and you can read the text they are pointing to. Finally, large bandwidths allow us to reassess the economics of connection-oriented protocols. The connectionless packet protocols of the Internet are very vulnerable to denial of service attacks. For applications where these attacks are undesirable we may be able to use GMPLS controlling optical switches to give dedicated bandwidth on demand between end-points.
What could you do with fat fiber? (Linux Journal) Posted Aug 4, 2007 14:57 UTC (Sat) by mbottrell (guest, #43008) [Link] Simple: Larger porn repositories. ;-)
Sadly ... and YouTube and video piracy Posted Aug 4, 2007 19:59 UTC (Sat) by AnswerGuy (guest, #1256) [Link] Sadly mostly the bandwidth advanced will be used by further transitionof pornography to higher resolution, full-screen, full-speed video formats (sucked up by such things, so to speak). Also there will be lots more YouTube style "pornography" (since quite a bit of the YouTube content is already akin to exhibitionist masturbation (sometimes not just figuratively). And, lastly there will be a greater explosion of video as well as music piracy. Naturally the rest would be consumed by spam and virus/worm (botnet) scanning and attack "radiation."
(From what I hear the "background radiation of spam and worm scanning
While I'm opposed to DRM, proprietary codecs, and the erosion of "fair
What would *I* do with ~54Mbps of low latency, stable bandwidth?
I personally would feel much more comfortable offering good mirrors,
JimD
I've had 100Mbps at home four years ago Posted Aug 4, 2007 21:36 UTC (Sat) by gvy (guest, #11981) [Link] In fact, the homenet's still 100Mbps (copper in-house, fiber between houses) but I'm on a "2Mbps unlimited" plan. Rather awkward story but in Ukraine, it was possible to run 100Mpbs UA-IX traffic very cheaply or flat rate some six years ago or so; since then, some providers carteled to charge for what they technically have very cheaply.
But I digress... when one has e.g. 100Mbps unlimited at home, it becomes very convenient to store stuff on public servers at a colocation and categorize it there for common convenience (I almost gave up storing ISO images at home since fetching CD one from ftp.linux.kiev.ua would take three minutes on average). It's also quite convenient to just ssh -X into the office, albeit the limiting side would be DSL there. :)
I've had 100Mbps at home four years ago Posted Aug 4, 2007 23:31 UTC (Sat) by khim (subscriber, #9252) [Link] Rather awkward story but in Ukraine, it was possible to run 100Mpbs UA-IX traffic very cheaply or flat rate some six years ago or so; since then, some providers carteled to charge for what they technically have very cheaply. Yup. They decided that it just stupid to sell the traffic below cost. Have you ever thought about the cost of this "very cheap" traffic ? No ? It's not very easy (there are a lot of factors involved) but you can start with price of CISCO router and capacity of said router. Today you can buy 7600 series with 720 Gbps backplane (the fastest available) for ~$15000-20000. It can handle something like 300-400 100Mbit connections (theoretical peak is 720, but it's unachievable and you need spare capacity as well). And if your network includes more then 300-400 end-users you'll need more then 1 router in chain - probably 3 to 5. Thus just 80-100 such 100Mbit users per router. And if you'll include other hardware required (repeaters, UPS, etc) you'll be forced to multiply this figure by 3 or 4. Since router can work for 3-4 years as well we can use the simple approximation: one router supports 100 end-users for a year and then it's replaced. So... 100Mbit end-user will cost ~$200 per year! That's just the hardware in traffic exchange point! No wonder they decided that 100Mbit for end-user is too expensive to offer. I think with current generation of technology you can expect to have 10-15Mbit (or alternatively 100Mbit connection with no more then 10% average utilization), anything else is just too expensive...
Oops. My mistake. Posted Aug 4, 2007 23:50 UTC (Sat) by khim (subscriber, #9252) [Link] Of course I did a mistake in calculation: I've calculated price for 1Gbit client, not 100Mbit client. 100Mbit client will cost just $20/year... still sizable sum if you'll recall that it's just for a single exchange point hardware, not for the whole infrastructure!
I've had 100Mbps at home four years ago Posted Aug 5, 2007 11:15 UTC (Sun) by MathFox (guest, #6104) [Link] You didn't take an "overbooking" factor into account. I have a 6 Mbit ADSL but on average use ~1% of it; my ISP can (and does) multiplex more than 166 ADSL subscribers on a 1 Gbit link; I guess they use a factor of 20 and expect to feed 3300 ADSL users over that gigabit link.In a similar vein one can expect a 720 Gbps router to support 360.000 users at a nominal 100 Mbps with a factor 50 overbooking. Redundant head end routers come under $1/subscriber.
The biggest cost is to get the cat5 cable to all homes; even in a residential area one should count on one or two person-days of work per home to get the cable from home to "local concentrator"; this is where the cable companies and telcos have an advantage. They have the cables and only need to make a (relatively) small investment to make them fit for internet use.
I've had 100Mbps at home four years ago Posted Aug 5, 2007 12:23 UTC (Sun) by khim (subscriber, #9252) [Link] You didn't take an "overbooking" factor into account. Have you even read other responses ? Most responses for a question "what will you do with the fat fiber" are variations of "I'll use it, of course" - and that means problems with overbooking... When ТоÑка.Ru first introduced "unlimited" ADSL access is Moscow (few years back) they offered 128Kbit connection but added note that "if you'll use more then 20Gb per month you'll be punished" - there was talk about of this limitation is a big deception or mere annoyance. Note: that's "factor 2 overbooking"!!! Not 20 or 50... Even today typical usage is 30-40% (Bittorrent, eMule, etc) in a lot of cases - even with 5-15Mbit connection! In a similar vein one can expect a 720 Gbps router to support 360.000 users at a nominal 100 Mbps with a factor 50 overbooking. As I've said above: either you offer "100% capacity" and then somehow try to force everyone to use just 1-2% of that (and hear complains from users who do expect to use 50-60% of offered bandwidth) or you offer "1% capacity" and can handle load just fine. Actually most ISPs now offer you choice: either you buy 1Mbit to 15Mbit unlimited (some ISPs only offer speeds up to 5-7Mbit because faster ones are too expensive and are not used much) or you buy 100Mbit (or sometimes even 1Gbit) connection - but then you have traffic limitation (usually 10-20Gbytes/month, but something 100-500Gbytes/month). Of course usually users expect to use sizable portion of offered bandwidth, not mere 1-2% (because there are slower tariffs and if you don't need so much traffic why will you spend money for faster tariff?). To offer you "100% of fiber" (not matter what does it mean) is too expensive without overbooking - and what will you do if you expect to use overbooking to solve "excess copper/silicon cost problem" while end-users expect to use 100% of offered channel ? With "factor 50 overbooking" just 2% of "bad users" can screw you... and then both these users and all other users will complain that everything is slow, you are not providing what you've promised, etc...
I've had 100Mbps at home four years ago Posted Aug 6, 2007 15:45 UTC (Mon) by peace (guest, #10016) [Link] How are you even going to fill a 100Mbit line to 40% capacity for the month? Are you going to copy the Internet to /dev/null? Even if you could utilize 40%, how many home users even use what they have now? You'd be whats called a "corner case". 100Mbit would just be marketing fluff for most users.
Would I like to get my isos 100 times faster? Of course! But after 3 minutes of downloading I'd be off the network to burn them.
-peace
I've had 100Mbps at home four years ago Posted Aug 7, 2007 7:33 UTC (Tue) by khim (subscriber, #9252) [Link] How are you even going to fill a 100Mbit line to 40% capacity for the month? With Bittorrent ? It's easy: go to tracker, grab 100-150GB of stuff and keep you client alive. That's all. Would I like to get my isos 100 times faster? Of course! But after 3 minutes of downloading I'd be off the network to burn them. And that's where things are becoming ugly: 100Mbit connection will not give you significant speedup. FTP servers will become clogged and that will be it. P2P (bittorrent) can give you such speed IF someone will keep the herd alive and well. And alive ad well here will need 40% of bandwidth usually :-) Even if you could utilize 40%, how many home users even use what they have now? Depends. Unfortunately the fact that just 5% users use 40-60% of channel globally will not save you. If 20-30% users connected to one fiber line will use the whole capacity of their connections - you'll be screwed. And it's not hard to imagine something like this. How many users will care if their 802.11n is not secure enough if ISP will supply unlimited 100Mbit connection ? It can easily lead to 90% usage before owner will think that he must do something! 802.11n is future, 802.11g is today: toys like Nintendo DS can not cope with WPA, but WEP newtwork today can be considered wide-open. Again: you fail to see that with sizable overbooking (20-50 mentioned above) tiny percentage of users can screw everyone. And small overbooking (2-3 times or so) the equation does not change materially: it's still way to expensive to offer "100% capacity"...
Bit Torrent Posted Aug 10, 2007 21:55 UTC (Fri) by ttfkam (guest, #29791) [Link] ...is the solution, not the problem. Or rather P2P in general is. Can we all download ISO images off of a remote FTP site simultaneously at 100Mb/sec? Most likely no for obvious reasons; however, we can absolutely download a torrent file from that server and download at 100Mb/sec through P2P.
How?
If 30 people on the same Verizon segment, let's say in the same zip code in southern California, connect to each other as peers, how does it harm the backbone significantly? Certainly their connection to networks outside of Verizon's network will be at reduced speed, but they will likely be getting different pieces of the same pie. All pieces that each have download would be able to be shared very rapidly.
Isn't that one of the points of Bit Torrent: to download rapidly from close, fast peers when possible, reducing the strain on central servers?
In addition, if people have a larger pipe especially on the upload end, they will be more likely the leave their clients running. More clients translates into a greater likelihood that what you're downloading is located on a network near you. As common as it may be to make the association, I think we will all be better off making a strong distinction between P2P and copyright violations. They may have been correlative, but they are not mutually causative.
The problem to network bandwidth problems is not P2P. The solution is P2P.
I've had 100Mbps at home four years ago Posted Aug 12, 2007 20:53 UTC (Sun) by silpol (guest, #46731) [Link] oh, now I see lovely mental fixation of ex-compatriot ISPs - *charging for breathing, per minute or per volume, but never flat rate*, it's in DNA :) I've been talking to so many ISP guys in Russia, and when they hear "flat rate is the only fair model", they all go crazy and start to explain me with dropping saliva, how dead-wrong I am, and how people like me should be kept isolated from society, and how it is 100% wrong to sell flat rate, and why people should be charged... You can't fix greedy brain :)
And... the most amusing part was when now-ex-ISP guy was pissed off of absence flat rate in available tariff plans in his part of Moscow, just few years later after he had quit his job at ISP, tho just few years back he was merely shouting on me, while I was calling for opening competition among ISPs and last-mile unbundled... :)
and, please, don't sell us cheap story about costs vs. availability - margin is not about upgrade of infra in Russia - once in place, it is forgotten until it really kicks in the ass. money goes to pockets and bank accounts - that's it ;)
lucky I am, I live in country where flat rate on wired lines is the only way to survive - thanks to unbundled last mile and strong control from govt over monopoly in this area :)
What could you do with fat fiber? (Linux Journal) Posted Aug 6, 2007 16:25 UTC (Mon) by Baylink (subscriber, #755) [Link] We can't do crap with it, because getting incumbent providers to provide symmetrical bandwidth is unlikely to happen.
Business services are provisioned to a higher level of service, which is why they cost more. Lots of people who need the bandwidth but don't need the SLA will abandon business for 'residential' service levels if they can, and the economics of the whole thing go to hell.
It's just like Canadian drugs. Does anyone at all think that if the pharma companies are forced to adapt US prices to match, that the results will *not* be Canadian prices going *up*?
What could you do with fat fiber? (Linux Journal) Posted Aug 7, 2007 7:40 UTC (Tue) by khim (subscriber, #9252) [Link] We can't do crap with it, because getting incumbent providers to provide symmetrical bandwidth is unlikely to happen. Most fiber-using providers offer symmetrical bandwidth in my region. Except ones who play games like overbooking and QOS management to offer more then their backbone can actually carry. Lots of people who need the bandwidth but don't need the SLA will abandon business for 'residential' service levels if they can, and the economics of the whole thing go to hell. If by "economics of the whole thing" you mean "business pay more and cover losses evoked by residents" then this model was never sustainable.
What could you do with fat fiber? (Linux Journal) Posted Aug 12, 2007 21:05 UTC (Sun) by silpol (guest, #46731) [Link] Before you've decided what you would do with extra bandwidth, think about actual level of installation-and-customer service within Verizon FiOS service - Om Malik's Gigaom article "Million FiOS customers and counting" has nice link in the end, "Verizon tech sets fire to home - accidentally", read it and think again...
What could you do with fat fiber? (Linux Journal) Posted Aug 16, 2007 20:07 UTC (Thu) by anton (guest, #25547) [Link] What could we, as Linux developers and users, do with fiber to our homes and businesses?"From what I read, large parts of (former) eastern Germany have fiber to the home, because that's what looked like the right thing to do when they modernized the phone system after the German unification. So what do they do? They get no ADSL service (that needs copper for some reason), so they get 64kbit/s pay-per-minute ISDN, or something like that; at least that used to be the situation two years or so ago. So while fiber to the home is nice in theory, apparently it does not have the network effects yet to make it an economical high-bandwidth internet connection for home users.
|
|||||||||||||