32 thoughts on “What speed is really Broadband for you?”

  1. Whether things in the UK are drastically different to the US – I don’t know, but I’d consider 512Kbp/s to be broadband, though with anything from 200kbp/s to 25Mbp/s being lumped together as one and the same, maybe the solution is for another level of distinction, rather than changing the one we already have.

  2. I am of the opinion that focusing just on downlink, nominal bandwidth is misplaced. Instead the measurement should be based on some statistical measure – something like sustained x Mbps, y% of the time and one should insist on comparably improved capacity on the uplink. Otherwise, we are focusing on an empty promise for “web 1.0” service paradigm.

  3. I don’t think the band will really ever be broad enough. The fatter the pipe gets the more ways we find to stuff it.

    Especially when you take things like Joost and Skype who use ‘SuperNode’ peering technologies, it will never be enough.

    But thats part of the fun of this whole industry, we’re never satisfied.

  4. Well, obviously this is a moving target. 200 kbps was broadband 10 years ago — there wasn’t much web and a 56 kbps leased line could handle a lot of e-mail. Today, there are few applications that need more than 1 Mbps (high quality video-casting being pretty much the only one I can think of). Clearly, as demand for video grows, the target for broadband will move to > 10 Mbps. One thing that should be clear is that when we are talking broadband speeds, we are talking average not peak. I could care less if something is capable of 2 Mbps, if I never get better than 700 kbps (I’m talking to you, 3G).

  5. 2 Mbps or more downstream, 1 Mbps or more uplink. Any clue about what %age of the broadband users in general get that kind of bandwidth (i.e. 2 Mbps or above)? I would suspect it would be in the single digits.

  6. Keep in mind the definition is all about what the minimum standard has to be for truth-in-advertising purposes, not what we’d like to have in an ideal world. I am quite happy with defining broadband to be 2Mbps, as long as that is symmetrical. In any case, due to something known as TCP slow-start, only very lengthy file transfers ramp up to use the full bandwidth available, and differences in downstream bandwidth are mostly irrelevant for basic tasks like browsing.

    The discussions in Congress did not at all raise the subject of upstream bandwidth. The lousy upstream speeds of ADSL and cable are what gets me most riled. Until we have decent minimum bandwidth, it won’t be possible to run services like network backups.

    Internet transit is symmetrical, so there is no cost reason for last-mile nadnwidth not to be as well. The main reason is that cable providers often don’t have enough free channels available in their cable plants, and telcos don’t want to cannibalize their lucrative T-1 business (even when, as in France 10 years ago, a “T1/E1” is actually a SDSL circuit).

  7. The good news is that the FCC is finally realizing it can’t live by its own numbers anymore (in the past the FCC would count an entire zip code as having “broadband” if one household in the zip code had service). The commission is on a similar tack to improve its reporting structures, but my question is — couldn’t the community at large do this better, a google maps mashup where we all submit our bb speeds?

  8. Great idea about the Google mashup, Paul. The best implementation could be done by Google itself. People don’t even need to submit anthing! Google has the technology to know our connection speeds. For all we know, they are probably already recording that information about everyone who visits their site. My Google Analystics report tells me the approx. speed of my website visitors. So, it’s not a big deal for them to show something like the maps mashup you speak of. Who knows, they might already have it internally!

  9. You know, “broadband” is a relative concept. Most applications I use will suck at 56 kbps, but they would do just fine at say, 256kbps. YouTube won’t, but it’s not like, mission critical, is it?

    Looking at the transfer rate of most of the data I download to my computer, 1.5 Mbps–and generally, well under 1 Mbps–would be plenty. The limiting issue is usually how much upstream bandwidth has been allocated to the server that is feeding data to me. My ISP provides me a connection with up to 6 Mbps, and my bandwidth speed tests tell me that I’m consistently getting well over 2 Mbps. But the limiting factor is almost never on the receiving end. Consequently, I’m a bit puzzled by why so many people are obsessed with how much downstream bandwidth they have. I find issues like latency to be far more frustrating than raw throughput.

    For what little it’s worth…

  10. Considering the telcos promise, promise, promise and never follow through on rural and lower income markets – I think it’s time to force them to offer 10Mbits minimu nationwide or lose their USF fund shares It also SHOULD NOT be up to the FCC anymore – they are worthless. (see below)

    The Federal Communications Commission
    [February 1999] Chairman William Kennard obtained a commitment from AT&T to extend advanced communications services and equipment to rural and remote areas as a condition of regulatory approval for its TCI buyout. However, AT&T’s promise to deploy broadband services to high-cost areas remains ultimately unenforceable.

    [September 1999] The FCC raised the spectrum cap to 55 MHz in rural areas to spur deployment of wireless broadband access in under served remote and rural communities. But wireless companies have been slow to jump into serving less profitable rural markets.

    [October 1999] The FCC made DSL deployment to rural areas one of 30 conditions applied to its approval of the SBC-Ameritech merger. Yet another ultimately unenforceable proviso from the FCC’s much maligned merger-review process.

    [November 1999] The FCC adopted rules to promote competition for advanced services by directing local telephone companies to share their telephone lines with providers of high speed Internet access and data services. The line sharing order was designed to flood the market with DSL services from either incumbent or competitive carriers. Naturally, DSL providers cherry-picked urban market deployments to reap maximum profits.


  11. I would say that 2 Mbps is broadband for up. My download speed over the past couple years has gone from 3 Mbps to 13 Mbps down but I pay extra to increase my upload from 0.5 Mbps to 1 Mbps. If you let people download at 1000 Mbps it won’t make as much difference as increased upload speed would make for the internet as user driven content is more popular will will continue to be even more so.

  12. I am impressed with your readers. Most of them understand that the “published” download number it not as meaningful as the upstream network and latency. I just got through watching the 2.5 hour replay of the congressional hearing and all they want to do in mug for the camera and score political points. I can’t speak to the throughput experience in Europe, but I spend a lot of time in Japan and it is VERY subjective. Sometimes it is better than the US, but it is also worse.

  13. I agree with Stuart Frisby. 512kbps up and down seems like broadband to me. Maybe it’s cuz I was raised on 1200 bps dialup. 512 should be enough for any non-real-time application, plus downloading large media files.

    I’m not complaining about my 15 Mbps, but the past ten years have just been icing.

  14. Maybe the term broadband without a qualifier is obsolete — what are you trying to do — then I’ll tell you if your pipe is big enough.

  15. There’s also a big difference between published cable “best possible download– but will vary if lots of people are on at the same time” and FIOS (and presumably other fiber services), where you always get the published number. (Of course, the latter is also because they have so much extra bandwidth that the only reason you’re not getting more is that they’re capping it.)

  16. Very funny. I always thought that fast or even very fast broadband was standard in the US and that Europe was far behind. Now you refer to Europe in your poll as the continent with eminent fast internet connection.

    I would say that the average speed in Europe differs a lot from country to country. In Scandinavia and France 16+ Mbps is not uncommon, whereas Germany is more in the middle with average 2 till 6 Mbps. In Eastern and South-Eastern Europe the connections are usually much slower. And: The broadband penetration in almost all parts of Europe (except Scandinavia) is lower than in the US.

  17. All networks are shared at some point – even FIOS. The weak points for FIOS are link speeds in the aggregation network and the attachment to the core. See what happens when you get 10k people running Joost. Think of sewer pipes. Running a bigger pipe to your home only helps if the rest of the system is engineered to handle the flow from all the homes and businesses.

  18. BTW – Just for accuracy in reporting, the FCC 200kbps definition is for “High-speed” not “broadband”. Most of the members of congress were very careful in using the term “High-speed”, as well. Unfortunately, most of the people reporting on this hearing obviously didn’t actually attend or listen the hearing. You can download the WMV file on the House website.

  19. Kate – “The last mile” (the line up to your house) has typically always been considered the slowest part of the network, according to my understanding.

  20. Running a bigger pipe to your home only helps if the rest of the system is engineered to handle the flow from all the homes and businesses.

    Yes, so there’s a point where FIOS speeds don’t help because the rest of the system is too slow. That’s why, e.g., it makes a big difference for me in speed tests whether I connect to Speakeasy’s DC server or somewhere farther away. Speakeasy’s DC server always gives me a 30 Mbps down / 5 Mbps up connection, but places with more and different hops between me are much slower.

    However, there’s little that we can do to control the rest of the network. It’s still a big change to move the bottleneck from the last mile to somewhere else, more reminiscent of being in college.

    From a practical standpoint, yes, there are diminishing returns to getting more bandwidth if the rest of the network and the backbone doesn’t and they can’t handle it. But I’d still rather have “enough” bandwidth and know that most slowdowns aren’t the fault of my connection.

  21. The last mile is the “slowest part of the network”??? Do you really think FIOS can guarantee all its subscribers 50Mbps throughput to Youtube at the same time? The sum total of the access network is the fastest part of the residential network. Every link upstream gets progressively slower (in aggregate). The Internet core is in great shape because it is statistically multiplexed over so many users and CDN’s offer offload. It’s the metro networks that won’t be able to keep up, if everyone starts pumping 30 minute TV shows to each other.

    That’s the difference between residential and business service. Business service is usually “committed rated” and you pay a lot more for lower access speed numbers, but you can use all of what you think you are paying for. Residential service (even FIOS) is massively under provisioned. As bad as the under provisioning is in the US, Europe is worse and Asia is much worse.

    It’s like insurance. They are betting everyone won’t submit a big claim in the same year. If there is a major event (like P2P video, which can cause network storms), insurance companies are in trouble.

  22. From what I gather, typical contention ratios for residential broadband are around 50:1. If all subscribers are running full bore, then you could theoretically expect 1/50th of your access speed. Actual application performance would probably be worse, since network efficiency goes down as contention increases, especially with greedy application like BitTorrent that open multiple simultaneous TCP connections or real time apps that use UDP without flow control/congestion avoidance.

  23. Just be grateful you don’t live in the broadband backwater of the developed world: Australia. According to one study we rank 26th out of 27 developed countries when it comes to access speed.
    “A report from New Zealand firm Wairua Consulting published in 2006 places Australia 25th out of 26 developed nations in terms of average download speeds for DSL broadband, just above Mexico but behind Slovakia, Poland and the Czech Republic.”

  24. Hey thank your lucky stars you aren’t living in South Africa, where 4Mb/s is a luxury and until recently was unofficial and only for “test” purposes, and the average user has between a mere 384Kb/s and 1Mb/s. Oh and add to that the fact that no matter what speed you have, your usage is capped, generally at 3GB. It’s a joke, just not a very funny one.

  25. To me broadband would be at minumum symmetric 2Mbps line, preferable 10Mbps, and I do not much care whether it is DSL, Fiber or Cable as long as it is symmetric and have low latency and a guaranted minimum speed. Of course there needs to be flatrate!

    I live in Sweden where the so called broadband coverage is rather high. We do have flat rate 🙂 The thing is that there is no problem to get 24Mbps ADSL (Asymmetric Digital Subscriber Line), the issue is that with 24 down and 1Mbps up the “pipe” will be filled with ack messages 🙁 I’ve had 512/512kbps, 1mbps/512kbps, 24/1Mbps on ADSL, 8/1 Mbps on cable. Those works pretty much the same way, although ADSL generally have been more stable and having lower latency. Just recently my brother got 24/2.3Mbps ADSL and finaly it was possible to use the download speed more efficiently (not only sending ack…).

    Someone said that the pipes will always be used and that is correct. Movieclips, Video On Demand (VoD) etc is getting increasingly poppular (as of today around 60% of the traffic on internet is video, by 2010 this figure is estemated to increase to 98%). In addition satellite services are starting to send in HD quality (some in 720p others 1080i) which will further increase the demeand on broadband capacity when more and more companies etc will start to compete with these satellite services….

    With services, such as the ones Google provides, where you work with “thin” browser clients, storing your data on virtual drives (located on Internet) etc the demand will just increase and increase.

    In addition, several companies starts to offer “Tripple play” solutions. That is; telephone, broadband, and TV via the same media, ADSL, DSL, Fiber, Cable.

    Imagine when 3G with flat rate will boom as well..

  26. Currently, I’m on FiOS at 30/5. I was on Verizon DSL at 3000/768 and have had Comcast HSI at 6000/384 (though PowerBoost “muddies” the bandwidth waters to a degree). I would call each of these options “broadband” as much because they are dedicated, 24/7 (effectively) connections. I think of broadband as something which is the opposite of dial-up, as something approaching true networking. It’s not so much the speed which concerns me as the availability; as long as the speed I get is close to the speed I pay for, it’ll be “broadband enough” for me. Customer-selectable tiers such as Verizon offers with FiOS and variable speed caps such as PowerBoost enhance customer options.

    The minimum speeds to be considered “broadband” should reflect a minimum activity. For most users, the minimum activities are e-mail, online bill paying, and the occasional shopping and general browsing; neither videos nor music come into play. As such, 768/128 is certainly enough for most users now and is a speed that would attract all dial-up users if the price is right–no more than twice what they now pay for dial-up. That’s the real point: how to get dial-up users off of dial-up. Since broadband penetration is not currently pervasive to the point of providing [potential] customers with more than one alternative, none of which might be cost-effective, many will simply stick with dial-up; they don’t need “broadband” at $30 to $40 per month, if they need it at all. For my purposes, however, 1500/512 would be sufficient more than 95% of the time. Since my “sweet-spot” for Internet access pricing is about $50, that’s about how much I spend, and I want to get the most for my money. Luckily, I live in a FiOS-served area, and the 30/5 tier costs $55 per month. And it’s definitely BROADBAND.

    Lastly, as has been stated, whatever is considered the minimum now will be different only two years from now, but this minimum is for most users and not the relatively few who want/need more bandwidth for music and video and downloading/uploading. The people for whom this minimum is designed aren’t even much concerned with this topic–they’re still on dial-up.

  27. What good is a minimum speed for someone who can’t get broadband? There’s sill large areas around here who’re stuck with 28.8 dial-up (which isn’t even that) because of where they live. They need to worry about availability first, then minimum speeds.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.