Slashdot Mirror
Does Faster Broadband Matter?
tsa writes "There is an interesting piece on Ars Technica discussing the implications of faster broadband services for the users, and for the internet as a whole. From the article: 'Most online activities, like standard websurfing, are not significantly sped up by high-bandwidth connections, and the few that are, such as downloading, are not typically time-sensitive anyway. Many service providers are starting to prioritize their own content at the expense of those from rivals. Many countries have started or are considering blocking Voice-over-IP (VOIP) traffic in order to protect the phone companies from competition.'" How does faster broadband actually impact your Net usage?
Can your eyes tell the difference between a web-page loading in one second or 0.27 seconds.
I guess if you only consider standard web browsing when considering if faster broadband matters, the answer is likely that it doesn't make much of a positive impact. At least two things that this fails to take into consideration though are:
1. There are far more applications today that can utilize the faster broadband, both upstream and downstream. For a few examples, consider P2P, VoIP, video streaming, etc.
2. Increasing broadband speeds and their adoption rate enables new applications tomorrow.
Give many people more bandwidth; they'll find a use for it. Feel free to replace "bandwidth" with just about anything and it likely would be true as well.
I'm a big tall mofo.
. . .don't download tv shows, run a web server from their closet, and download large ISOs of operating systems.
Huh, maybe you shouldn't ask this question on Slashdot.
Porn.
A monster ate my homework!
as someone who has 100mbit fiber to the home in Tokyo: Absofuckinglutely.
You lost me at the phrase "Internet blogger".
-William Brendel
Everytime new technology comes out, someone always says "Nobody needs that much memory", "What would ordinary people want to do with a computer?", etc...etc...but as we start to experince this new broadband boom, we'll see dozens of services that were just waiting to come out, Video On Demand rentals of HD Content, Full Stereo Phones, Video Phones (Instead of crappy webcam chats), and more I'm sure someone with more time will think of.
my isp (Verizon, which is the huge phone company here) is planning on converting all of the DSL lines to FIOS (fiber optic) to allow like 24mb speeds. they are doing this to offer cable TV as well as internet and phone service all through one handy dandy line. This will be great since there are no cable companies in the area so I have no cable TV but do have broadband internet. I say bring on faster speeds, they will bring me TV channels and allow my web/mail server to run alot faster.
As a gamer, you should be more concerned about latency than speed-- at least, if you play "twitch" games (read: FPS games), as opposed to MMORPGs.
With spending like this, exactly what are "conservatives" conserving?
Excuse me? Downloading... not time-sensitive? If downloading isn't time sensitive, I don't know what is. Even for leisurely things like movie trailers, I don't want to wait more than is necessary. For people who transfer large files as part of their job, download and upload time is even more important.
Small potatoes make the steak look bigger.
The definition of broadband is specific: Broadband in general refers to data transmission where multiple pieces of data are sent simultaneously to increase the effective rate of transmission. In network engineering this term is used for methods where two or more signals share a medium.
Marketing are to blame for the confusing usage, where broadband means "really fast". This means we can look forward to terms like "ultrabroadband", "superbroadband", "megabroadband" and "bukkakebroadband" in the future (where "bukkake", meaning "to splash" in Japanese, will refer to a newer form of "spread spectrum"). For proof that marketing is to blame, see this link above and look for "confusing".
http://www.thebricktestament.com/the_law/when_to_
NO.
Bandwidth speed does not matter -- latency is the key to a happy user. These two do NOT have to go hand in hand, though.
I started (back in the BBS days about 21 years ago) at the age of 30 with a 300 baud modem, and quickly jumped to a 1200 baud modem. I took in information quickly (of course, a young mind is a sponge). My phone bills were $300+ per month -- requiring me to work.
I transitioned to modem's fastest and then transitioned to ISDN. The ISDN's latency was intense -- everything was amazing, comparable to the few T1's I had worked with up to that point.
I was the first of a very select group of DSL (IDSL) testers in Illinois before it really hit. I believe Michigan had it first but I had a consistent 144kbps up/dn connection and it was QUICK. Not as snappy as the ISDN, but download speeds were over double. Web sites, though, were not as snappy.
I switched over to ADSL and the snappiness went down but the downloads went up. Then SDSL, then cable modem, to where I am today -- cell phone dial up.
I just switched to T-Mobile's EDGE network. I get a consistent 150kbps down and 40kbps up from my PDA/laptop bluetooth tethered to my t809 phone. The latency sucks. The bandwidth is just about perfect, though.
I still download, upload, blog, e-mail, browse, etc. I have access to a T1 (at a customer's office) and an OC3 (also at a customer's office). Even though my PDA and my laptop both support WiFi, I stay on my bluetooth 150kbps connection -- just to keep things simple and keep battery life UP.
I've spoken with users of all sorts -- laymen and power users -- and they all tend to agree. Faster response is better than faster downloads. This is untrue for the younger users with time on their hands: they NEED fast downloads for BitTorrent and porn. Once you become part of the grind, you want quality web views with quick response times. I've switched some clients from high bandwidth DSL to low bandwidth DSL that offered lower latencies. They're MUCH happier.
FWIW, the order of need in my life:
1. Be available everywhere (EDGE/GPRS is close)
2. Have a low latency (EDGE/GPRS does not have this)
3. Have a decent download speed (EDGE/GPRS has this)
4. Be priced in an unlimited transfer package (EDGE/GPRS has this)
The only thing my current connection needs is a better latency. This will come with time, I hope. As for VoIP and the like, who cares? My cell phone bill is around US$100 per month -- offering unlimited everything. This price will only go DOWN over time, so I believe the phone companies are too little, too late.
True that. At the very least, we could have some of the most efficiently running spyware around. It could eventually get so good that it reports on the porn you saw tomorrow.
When I tell the ladies about my fat pipe, they want to come over to my place and stay up all night long.
Downloading movies.
Faster speed means I CAN browse the internet. A large portion of the internet is becoming nearly unusable for dialup users, especially the ones that can only get 14.4kbps because the phone system hasn't been updated since Nixon was a president.
If you think education is expensive, you should try ignorance -- Derek Bok, president of Harvard
The writer of the article seems to be worried most about a "two-tiered Internet" and how the networks are looking to prioritize some traffic over others. I don't see what the big deal is -- prioritization has been built into the IP protocol for decades now. Most network operators, however, have ignored the priorities.
The main reason for this is that if they started accepting priorities from their customers or peer networks, then their customers and peer networks would all set their packets to the highest priority. The end result is that traffic would be routed the same. For QoS to be of any use, there has to be a reason for people not to use it. And, money is the best way of doing this -- if you want to limit the number of high-priority packets going across your network, charge the people who put them there more than if they put low-priority packets on.
Streaming media requires a different type of service than do web pages -- if your GIF logo takes an additional 100 ms to load, you probably won't notice. If, however, a chunk of your phone call takes an extra 100 ms, you will notice it.
The problem comes in when Internet Video becomes widespread, because its need for high bandwith will overwhelm the rest of the content on the network. Prioritization won't help because almost all of the traffic will be video.
The real reason for allowing prioritization is that network operators won't increase their bandwidth without it. Think about it -- why would your cable company spend a lot of money on its Internet service so somebody else can use the Internet service to compete the cable provider's pay-per-view service? The only way the cable company will do it is if they can get a cut of the action.
I work for an ISP in the U.S. We have provided DSL at link speed since initial offering five years ago. That means, if the link negotiated at 1M by 6M, that is what the customer got. We configure all customers at link speed. We also have wireless Internet connections. They are configured at link speed as well.
Our observation is . . . that the faster the customers go, the faster they get on the Internet, the faster they get their surfing done, the faster they get off. And, the proof is in the numbers. With a sample of 500 link speed customers linked at an average of 800kbps up and 5000kbps down, we use no more than 5000kbps of upstream bandwidth on average and 9000kbps at maximum.
And, we have played with the numbers. Slowing customers down to 2000kbps was completely un-noticed by the customers. But, the average and maximum upstream bandwidth rose slightly. Slowing the customers down to 1500kbps was noticed by a few customers. But, the average and maximum upsteam bandwidth rose by 30% respectively.
So, by the numbers, the article is right. Customers use about the same amount of network no matter what. It is a matter of convenience/efficiency for the provider to give the customer a faster pipe . . . for their own benefit.
Does this mean that everyone is being manipulated . . . sure . . . but, it isn't the fault of the network guys. Blame marketing . . . They are the folks who like to manipulate people.
It depends on your definition of fast. Most people equate fast to the amount of bandwidth they have. The fact is, most online games will not saturate your typical broadband connection. When it comes to online gaming, you really need low latency. It doesn't matter if you have 10Mb down and 1.5Mb up if you have 500ms latency!
The problem is that residential broadband service providers crank up the bandwidth but do not guarantee latency. Perhaps someday they will sell a product geared towards gaming with a latency SLA.
And 640K should be enough for anyone right?
How could anyone say that more bandwidth won't find applications? It's dumbfoundingly stupid.
On the other hand page loads are not really set by the connection speed. After about 40K per second it's the servers and the latency that sets the download speed. That's one reason why things like google's "secret" data-center-in-a-shipping-container project will be important to frontloading content closer to the destination.
We have yet to reach a point where one can replace a desktop with a thin client or dumb terminal. But Sun's sunray show this is indeed possible if you have enough bandwith for the video connection.
Outside of high performance LANs you can't do this. But with ubiquitous high speed connections of the future only a fool would actually want to own and maintain his own computer. It'll be a paradigm shift enabled by fast connections.
Some drink at the fountain of knowledge. Others just gargle.
And no, there's no simple way to reallocate frequencies and have more of it used for upstream capacity. Assignment of frequencies for cable video is a matter of federal regulation.
Just imagine if you replaced your current kitchen trashcan with one that is twice as large. You might think "I'll never need this large of a trashcan!", but how much you want to bet that a week later it's just as full as the old one, and you're saving time by only having to go to the curb half as often. Granted, the new trashbags cost more, and it gets smelly, and strange things start growing in the trashcan that you can't readily identify, but how is that really any different from an always-on broadband connection?
SLA?!? What are you smoking for $40 broadband?
What cable/DSL providers give you a service level agreement (SLA) where they guarentee and back financially their uptime/availability, let alone the speed of your connection. They all provide no remedy for downtime, no guarentees of bandwidth as it depends on your area and usage. Why would they guarentee latency that has so many additional factors including line quality, distance, and the routing equipment used.
You won't find an SLA on anything less than a ISDN/T1+ connection. Maybe some sort of corporate broadband does, but in my experience even $75-$150/mo 'business' broadband has no guarentees either.
-M
when you see the word 'Linux', drink!
150 years ago, most people did not have running water. If you wanted to know all the benefits of running water would you ask people without or with it?
If all you want is email and browsing you can get by with a modem. All you have to do is turn off Flash and other crappy plugins and get a half decent browser that let's you block images from ad servers. I've done it and shared the line with my wife and the "normal" use worked just fine. Getting pdfs and other large files sucked life, but you could do that at night with a good download program.
GNU/Linux, with user driven development, is cutting edge and giving people exactly what they want from their computers. People want to share their pictures and dreams with family, friends and others interested. Blogging is now one of the easiest ways to do that, but it's not much harder to do your own when a Mepis CD will auto install Apache with most of the extras. It's actually much easier to make an html photo album on your spare computer than it is to carefully select and upload them to some place that will load them with adverts and go away in a few years. Getting your software off the network via ISOs or automated update tools are exactly what users want as well. Automated downloads from Debian, unlike some updating "services", are unobtrusive and can be trusted to keep your computer working well. Amazingly enough, people also want their Dick Tracy video phone.
Contrary to all of the above, the FCC is happy granting monopolies to greedy morons. By some twisted logic, they think that a cable monopoly competing with a telco monopoly will provide "enough" competition for people to get what they want and the providers to profit "enough" to provide new services. The greedy morons have been proving them wrong for five years or so. I can compare At Home and my choice of DSL to today and it's not favorable at all. Services have dried up with choice and the extra money is being put into an "intigent" network that will make competition in the future even more difficult.
Five years ago, things were much better. For less money that I currently pay for cable, I had better bandwith and fewer restrictions. Today, I have a cable modem with port blocks and a 60KB/s upload crimp. At Home provided the same without restrictions at all and the service was reliable. It was also much easier to get a DSL line, that did not suck, from someone other than the local telco. Today, we have the local telco and the cable company working to penalize each other's packets and the technology, of course, will slow everything up.
Greed, in this case, has been very bad. It's eliminated the companies that provided services people want and rewarded the assholes.
Friends don't help friends install M$ junk.
I think the interesting question is not how much raw bandwidth is available to users, but whether the move to tiered service for content providers will will catch on. If it does, I think the internet as we know it is doomed.
Step 1. Major backbones provide tiered service offering lower latency and higher speeds to content providers who pay a surcharge. Everyone else is assured that their service will not be adversely affected because they have plenty of execess capacity.
Step 2. Major networks, studios, advertisers, software companies, and national magazines all sign up for prefered status with the backbone providers. Consumers sign up for broadband in droves so they can watch truly high quality streaming media from the major content providers.
Step 3. Excess capacity gets used up. Banwdith partition devoted to those paying for prefered status expands, bandwidth available for everyone else contracts.
Step 4. A consortium of SBC, MTV, Time-Warner, and Ticketmaster buys all the Internet backbones. Web 2.0 becomes Cable TV 2.0. Microsoft re-launches Blackbird. The rest of us go back to using dialup BBS systems over 56modems that are then transmitted over VOIP.
This writer's conclusions make no sense.
Sure, most users don't use their broadband to full capacity. There's a huge different between a backbone internet connection and a consumer grade line. The entire consumer broadband business model is built on the concept that giving a very large number of consumers high speed access will work if only a small number of those users are generating substaintial demand at any one time.
He also misses the fact that current providers have adopted the asymmetric line speed model in an attempt to curtail peer to peer and hosted content by consumers. This artifical cap will slowly erode, as we've seen in FTTH and some cable offering already.
Also overlooked are emerging trends in smart houses, automation, video monitoring and tele-presence, all of which assume the easy availability of cheap, fast consumer bandwidth at the core of their business model. Other applications, such a remote medical diagnostics and imaging will also generate more usage and will be encouraged by employers and medical providers.
The entire premise of this article is biased from the outset. It really seems like he wrote the entire item to support a preconceived conclusion. Or perhaps it's another case of the media intentinally stirring the pot...
What would Groucho do?
... queueing is. What I want isn't more bandwidth, it's QoS.
I have a cable modem (~4Mbps down, ~400kbps up) and I use it pretty heavily. I run a mail server and a web server, frequently use VNC when I'm away from home, VOIP when I'm at home and often have a bittorrent download running (usually getting some recent TV show), not to mention the normal surfing and downloading activity of a half-dozen computers.
My problem is that latency can get really bad for interactive usage when something else is sucking up a lot of the bandwidth. When someone is receiving or retrieving a big e-mail, for example, surfing can get annoyingly slow, remote telnet/SSH/VNC connections get unresponsive and VOIP becomes useless.
The problem is that one network connection may receive a burst of data that the ISP helpfully queues up for me, so they can keep my incoming pipe full. I also see problems when I saturate the outbound connection for a little while. It appears that they do a lot of outbound queueing as well. The symptom is that round-trip packet times across the cable modem link increase to upwards of _3000_ milliseconds.
I can use traffic shaping to prevent queuing at the ISP, but only by severely restricting my total bandwidth. It makes my VOIP smooth, at the expense of slowing down everything to about 1.5Mbps incoming and 200kbps outbound. For those who aren't familiar with it, traffic shaping basically involves using a router to prioritize and manage the network traffic.
Let me explain how it works (as I understand it, corrections and suggestions are welcome!):
Prioritization of outbound traffic is a no-brainer -- if the router has a VOIP packet, an SSH packet, an HTTP packet and a bittorrent packet all waiting to be sent, it should send them in that order. Management of outbound data volumes is a little less obvious, but still pretty simple: The router limits the rate at which it sends packets. It has very shallow queues and rapidly starts dropping packets which can't be sent without exceeding the specified maximum data rate.
Inbound traffic shaping is less obvious, but also works fairly well. It relies on the fact that every decent IP protocol is not only tolerant of dropped packets, but actually takes dropping of packets as a hint to self-tune. TCP is marvellously good at this. So inbound traffic shaping keeps track of the data that has arrived (both volume and type) and if a connection has exceeded the limit, the router drops the packets. It may seem wasteful to drop data that you have actually received, but doing it will cause the sending TCP stack to slow down the rate at which it transmits, resulting ultimately in a smooth, continuous flow of data at very close to the target rate. To prevent a big "stall" when the data rate crosses the threshold, Random Early Detection (RED) can be used. RED will randomly drop packets even before the maximum rate is reached, with the probability of a drop increasing as the rate approaches the maximum.
Ideally, I should be able to configure my shaper to limit incoming and outgoing data rates to just a little less than what my cable modem can handle, and that should ensure that my high-priority packets (like VOIP) always get through right away.
It doesn't work.
Why? Because the ISP does too much queueing, and does it with a straight FIFO... no prioritization. So while I actually can get a sustained download rate of 4Mbps, latency goes to hell in a hurry. At anything above about 2Mbps my latency goes through the roof and to reliably avoid queuing I have to keep the inbound rate at 1.5Mbps or below.
I understand why they do it... so they don't have to buy as much total bandwidth. Queueing allows the ISP to serve more customers for a given amount of bandwidth to the backbone (yeah, I know, it's not "a" backbone any more). It makes congestion on the ISP's network connection less apparent to the end-user. Suppose I'm doing a big download, sustaining the maximum data rate my cable modem c
Note to ACs: I usually delete AC replies without reading them. If you want to talk to me, log in.
most users will not max their line even once a day as they just web surf and only occaisionally download mp3's (not very big files) or a new program or update (moderate sized). Only a small % of users are downloading iso's or similar sized data sets on a daily/hourly basis. So in that regard, no it doesn't matter. But individuals dont care about the 'big picture' of the generally small time savings they would get over a year using say 10Mbs down vs 1Mb. Most people want their download to be as fast as possible when they need it. As to browsing, the biggest delay I find now is the serving of ads not content. So many pages refuse to load, or only display partially, while waiting for these bs ad servers to send their stuff.
Streaming Video
Video on demand over the internet will be HUGE. The time-to-DVD for hollywood films can go down to zero, if there is a world wide release in theaters and homes. Piracy would be greatly diminished if people could watch any movie without needing to store them for a small price.
Robo-Blogs of the world: UNITE!
This has been solved. Swedish company Northspark has developed and are now selling a hardware product that gives each consumer in a building 1000Mbit/s duplex (for clarity: yes, its 1Gb both ways) over existing CATV/Coax networks.
2 553
http://www.newsdesk.se/view_pressrelease.php?id=7
http://www.northspark.se/