Five alternatives to current NBN shaping models


opinion As I’ve previously written, the concept of shaping users’ broadband speeds once they’ve exceeded their monthly quota, as currently proposed by ISPs, is an unsavoury anachronism from the ADSL days which has been unthinkingly ported into the fibre world of the NBN. We don’t like it; so here’s five alternatives to current NBN shaping models.

If you look closely at the current NBN pricing plans proposed by the ISPs who have released them so far, it’s not hard to see why many people object to the whole concept of shaping so much. The top speeds available under the NBN fibre currently range up to 100Mbps, but if you exceed your monthly quota, the ISPs will throttle you down to 512kbps (Exetel), 256kbps (iiNet) and a miniscule 128kbps under the worst offenders Internode and Primus. The shaping speeds get better when you sign up for some business plans, but only slightly so.

Such shaping speeds might have been relatively acceptable when customers were using ADSL2+ connections, where speeds top out at 24Mbps (although most people can’t get more than 16Mbps or so), but cutting users’ speeds down from a potential 100Mbps to the worst case of 128kbps is just plain ridiculous.

There is simply no way, in 2011, that anyone who needs to access broadband from home for work, education or even participating in basic community activities would be able to get anything done with 128kbps, which is barely faster than a dial-up modem. It’s just not possible – and the situation is even worse when you consider a family household with a number of different individuals needing to access the Internet. You know, the kind of scenario that the Federal Government loves to hype up so much.

Because of this, all of the proposals which we’ve outlined below focus on allowing customers who are high downloaders to still achieve the outcomes which they want (a boatload of traffic at a reasonable price), while still allowing ISPs to, you know, make reasonable profits, which is why they exist.

1. Increase shaping speeds
This one seems like a no-brainer, and it’s one that we’re going to be campaigning for over the next few years, until shaping speeds higher than 512kbps are a reality.

The simple fact of the matter is that 256kbps and 512kbps shaping speeds were set at that level because those were the (completely artificial) speeds which Telstra set up ADSL at (along with 1.5Mbps) when it first started selling this type of broadband. When ADSL2+ was rolled out, the idea that customers who exceeded their quota would be shaped to ADSL1 speeds caught on, and it’s one that has persisted ever since. We’re not quite sure where the 128kbps number came from … ISDN perhaps? We just know we really don’t like it.

So we’d propose some new tiers, more appropriate for an NBN world:

  • Customers on 25Mbps plans should be shaped to 2Mbps
  • Customers on 50Mbps plans should be shaped to 5Mbps
  • Customers on 100Mbps plans should be shaped to 10Mbps

If you think these speeds are pretty high, consider that they maintain the same theory behind ADSL2+ shaping. They restrict customers on one technology, in this case fibre, to the speeds available on legacy technology — ADSL2+. And I will bet you a pretty penny that customers shaped to these speeds will still be pretty unhappy about it. In addition, if you think they’re too high, I’d be interested to know what you think appropriate shaping speeds should be — feel free to post your thoughts in the comments.

If you’re an ISP and think these speeds are too high, consider that your main concern here is going to be that customers are getting data quota free, while you’re paying for it. In that case, why not raise the issue in public and start a debate over it, as Internode chief Simon Hackett did with NBN pricing a few months back? The Federal Government, after all, wants fast speeds for everyone. Perhaps different ‘shaped’ pricing can be negotiated with NBN Co. There’s no reason the company should charge a flat rate for everything. Let’s get granular.

2. Only shape high bandwidth services

This one also seems like a no-brainer. The overwhelming majority of those who download huge amounts of data to their home and business connection are currently doing so through a very small set of applications. You know the culprits I’m talking about. YouTube, BitTorrent, gaming platforms such as Steam, Xbox Live and the PlayStation Network (not playing games, but downloading game files). For some people, Usenet newsgroups even comes into the picture.

If one connection is downloading a massive amount of data, why not block these specific platforms for a period, say a week, as defined by a concrete acceptable usage policy, or until the customers’ new quota cycle kicks in? Or even just rate-limit those specific services, so a customer can get access to them, but not at an immense cost to ISPs? After a few days, when the users’ profile reaches a more acceptable range, the services could open up again.

An important idea here is that these services are predominantly entertainment-focused. I see it as acceptable to limit customers’ usage of really high-bandwidth entertainment-focused platforms for a while if they’re using them dramatically more so than other customers (for example, downloading dozens of movies that nobody would actually have time to watch). Of course, customers should always have the option to completely unlimit their services by paying more. And ISPs should welcome that option as well — more money for ISPs is always good, from their point of view. The key factor here is choice.

3. Fine-grained control over quota
Right now, most ISPs just provide one big block of monthly download quota. If you go over the block, all you can do is upgrade your plan, or, in a few cases, add on temporary quota packs to get through the month. However, aren’t there a bunch of alternatives here?

Why set a monthly limit on quota usage? Why not set a quarterly, six monthly or even yearly quota, as is now common in the mobile industry? I just bought 12GB of quota for my iPad from Telstra, which expires after 12 months. Why can’t I do the same with a broadband plan from iiNet or Internode? Let’s say, several terabytes of quota,that I could use throughout the next year. If I ran out of quota, I could buy add-on packs.

You’d obviously need to bolster the amount of data contained in add-on packs. Internode currently offers add-on packs up to 50GB, for example. But this clearly needs to be bolstered dramatically — with recharge packs starting at 50GB, perhaps, and ranging up to several terabytes. The idea here is that ISPs are locking customers into just one way of paying for services, where they could be offering many options, to suit each customer. It’s a competitive market. Innovate.

4. Focus on the heavy users
ISPs have complete access to customers’ address data. They have to — to connect a broadband service to their premise. So why not divide up their customer base into regions, do away with monthly quotas entirely, raise basic prices to compensate, and merely shape those customers in each discrete region who, for example, represent the top five percent of downloaders?

When a customer hits the top five percent and has their connection shaped, the ISP should contact that customer to let them know, and offer remediation plans. Offer them extra quota during the middle of the night, highlight cheaper mirrored data repositories located in Australia instead of overseas, offer them special ‘high downloader’ plans that would cost more but wouldn’t get shaped, and so on.

What I’m saying here, is don’t penalise everyone for the high-downloading activities of a few. Uncap broadband for the masses (who mostly won’t use that much quota). And for the top five percent, try and squeeze more money out of them, rather than penalising them too far for what is, after all, simply using broadband for what it’s designed for. This is a ‘mainstream’-style approach which is fairly popular internationally, but not really well implemented in Australia just yet, although the likes of TPG and Dodo are trying, with their ‘unlimited’ plans.

5. Simply charge per usage
During the NBN trials in Tasmania, Exetel unveiled a revolutionary pricing plan, proposing to charge customers a small monthly fee of zero dollars for a 25Mbps connection, $15 for a 50Mbps and $25 for a 100Mbps connection. Usage of that connection would then be charged at a flat rate of $1 per gigabyte downloaded.

Although the ISP subsequently backed away from the model, this pricing plan still makes tremendous sense, as it’s precisely the way that NBN Co will be charging ISPs for wholesale access to its network — a set monthly fee per customer, known as the Access Virtual Circuit, plus charges for Connectivity Virtual Circuit (CVC) backhaul charges.

Why shouldn’t customers be charged the exact same way? This basic system would see customers charged for the actual data they consumed, no matter if it was 25GB a month or 5TB. There’s no need to shape customers — you just pay for what you use. Perhaps as you get towards the high end, customers would receive a discount on the charge per gigabyte, to account for volume.

To sum up, I know that I haven’t worked through these models in detail in this article. To be honest, I do not have the technical knowledge or insight into NBN and ISP pricing structures needed to be able to discuss pricing in an NBN world at the level of detail which many of you would like. I am forced to leave that sort of detailed discussion to the likes of ISP chiefs such as Simon Hackett or Michael Malone, who have a much better understanding of the inputs which go into their business costs.

However, what I’ve tried to do with this article is highlight some alternatives to current ISP NBN shaping models which I think might form a starting point for discussion. I sat down with a mind map this morning and kicked around a dozen or so wacky ideas which could help to change the way that broadband shaping is done in an NBN world. Most of the ideas were crap. But at least I was thinking outside the box about the issue — which is more than most ISPs can say at the moment. And I’m hoping many readers and ISP executives will jump in and contribute more details as to how some of these ideas could work in practice.

Because I can tell you this. If I was paying for a 100Mbps fibre broadband connection to my house, and my ISP rate-limited me down to 128kbps, I wouldn’t just be hopping mad … I would be livid. Once the pipe is turned on, I don’t want it to get turned down. I want that NBN goodness to come gushing out of the tap whenever I want it. And ISPs need to realise that fact and start structuring their plans accordingly, instead of just copying them across from legacy ADSL frameworks, which, in 2015 … will look incredibly dated.


    • That seems fine to me. A bit on the harsh-side, but still fast enough to access basic Internet functionality. The whole idea of shaping is to essentially cut-off users who exceed their limits, but to give them access to basic functionality (email, simple web browsing) until their next month rolls around.

      If you are shaped more than once in a blue moon, then you are on the wrong plan. If you are one of the class of users who is shaped regularly, then your ISP is making a loss on you, so they will be happy if you leave. If shaping isn’t a giant pain in the backside for users, then there is no incentive for them to change to a sustainable plan. It should be as painful as possible.

      Other options mentioned in the article, such as yearly data quotas and $0 plans make no financial sense. The ISP is paying $27+ just to have you connected to the NBNCo (before you even start using the internet). So why would they adopt a business model where they may not even make back that minimal price per month?

      Monthly charging based on data is the way to go, because ISPs pay AVC and CVC costs to NBNCo on a monthly basis for their customers. Charging customers based on these costs just makes sense – any other option is a risky proposition.

      And shaping works. Provided it is pretty painful. If you hit your quota, you are already losing your ISP money (prices are set based on the assumption that users will consume on average between 30% and 75% of their quota, depending on the “plan tier” in question – lower percentage usage is generally the norm on lower tiers). Why should users be encouraged to keep going above and beyond 100%?

      • You don’t get it I dont care how old you are or what you do you simply do not get it.

        i hate dinosaurs like yourself that accept the rort of being slowed from 100mbit to 64kbit. Its just not right.

        it should be 100mbit and no shaping at all with unlimited quotas.

  1. The sensible thing to do is choose a plan where you dont become shaped.
    If you are being shaped regularly, then you are on the wrong plan.
    If you need extra for a month, then buy a datablock.

    • Two problems:

      1. If you download more than 1TB a month, there is no plan to suit your needs right now. They all top out at 1TB.
      2. Data blocks are prohibitively expensive. $50 for 50GB through Internode? No thanks.

      • 1. Telstra used to offer true unlimited business ADSL for about $300 per month. Have they stopped?

        2. Exetel charging $1 per gigabyte to everyone is “revolutionary” but Internode charging the same to people who suddenly want 50GB more than they planned for in a month is “prohibitively expensive”? What?

        • I think Exetel’s $1 per gigabyte was too expensive — I was referring to the model rather than the price. I like that Internode does data blocks, but $1 per gigabyte, in the current ADSL schema, and even in an NBN world, seems too expensive.

          You can get a 500GB/500GB NBN plan from iiNet for $99.95, for example.

          • What would the average iiNet user on that plan use per month though? 25gb? 50gb? 100gb? 200gb even?

            Once you start looking at using data blocks you introduce a problem for the ISP that their customers are going to stop buying quota they don’t actually use, and just rely on data blocks to top up their quota when they go over instead.

            This is arguably a good thing, but either the monthly prices need to go up, or the data blocks need to be priced high enough to discourage regular use, as like many industries, ISPs do rely on overselling.

      • If you are downloading more than 1TB/month, then can I suggest two connections?
        1 at 12/1Mbps for P2P and one for work that requires the speed?

        • Whats the point in that?

          It’s the NBN, why would you require 2 connections for it to do what you want?

          • Because you are buying from a residential service provider. Go to a business service provider if you need services that are not normally used by most residences.

            eg, If you go through 50 litres of milk a week, should you be complaining that Coles/Woolworths doesn’t sell milk in 50-litre bottles?

      • Buy 2 or more plans, from 2 different ISPs if necessary.

        Remember there are 4 data sockets on the ONT

  2. I disagree, Renai. The point of shaping *is* to frustrate you. That way, you pay more money (buy a datablock). The other problem is that ISPs still have to pay for the data you use, whether you’re paying for it or not. At 256kbps, the amount they have to pay is (comparatively) minimal, since the amount you can download is also minimal. Don’t like it? Get a bigger plan.

    To me the issue (and I believe you’ve written about this before) is that at 100Mbps, it takes you just over a minute to download a GB. That means it’s going to take 1500 minutes (or thereabouts) to download a TB. Just over a day. Now, one hopes I’m not going to max out my connection all of the time, but it’s still very conceivable that I’ll hit that 1TB mark. But then where do I go? I would have to pay (exorbitant) fees for datablocks. In this way, I really like the ‘pay per GB model’. $1 per GB seems a touch on the large side though.

    (Well aware my maths is probably a bit off. Shh.)

    • Well, part of what I am calling for is more datablocks (only a few ISPs offer them) and at better prices. 50GB for $50 through Internode is not competitive in an NBN world … you could go through 50GB in a few hours, as you pointed out.

      I agree highly with your second paragraph, regarding where you go after you hit 1TB etc. Right now, no ISP offers a > terabyte plan.

      • Renai, you and other journalists don’t seem to get it. The NBN is not equal to the Internet, and isn’t going to make the Internet cheaper – in fact it is going to make it more expensive for ISPs to deliver.

        The NBN is replacing the access network (i.e. the ADSL network), nothing more. To deliver Internet services, ISPs also have to buy the devices at the edge of the access network (BNGs), Internet bandwidth, pay their helpdesk staff, pay electricity costs etc. The NBN is only replacing one of those components. While the NBN might be making the access network bandwidth cheaper, it is not making any of the other costs significantly cheaper. It’ll actually significantly increase the costs of the BNGs and Internet bandwidth, because people with NBN access speeds (such as the 100 Mbps you mention), will expect faster Internet (i.e. like you are).

        ISPs aren’t going to deliver radically cheaper Internet access over the NBN because they can’t. Average customers don’t want to pay much more for their Internet than around $50 per month, and only *one* of the Internet delivery costs is being made cheaper by the NBN.

        The only way for the “NBN” to make Internet access radically cheaper and faster is to make *all* of the ISP costs radically cheaper. The government could do that by also handing over billions to ISPs to subsidise those other costs, but they’re not, and even that approach will have a time limit.

        • I couldn’t agree with this more. In fact, the pricing model offered by Telstra Wholesale is that of bandwidth, not data usage. ISPs pay for a big fat pipe from Telstra, and can utilise it as much as they want.

          The ISPs meter the data because the real cost is in transit, which is in fact measured in data. The two parties pay for either direction, and if the amount of data transferred is symmetrical, the money that changes hands is effectively equal. But what ISP (well, residential, at least, which is what most people here are talking about) would have symmetrical transit — it would be very biased towards the downloads.

          So that’s how ISPs can offer unmetered data. Because it doesn’t involve any transit. NBN still uses transit, which is why the pricing model must remain exactly the same.

          People are expecting something different, but really the current pricing model is not directly derived from the variable that is changing here.

          Apologies about posting anonymously.

          • An interesting theory, but completely wrong.

            The cost is bandwidth – it has always been bandwidth. ISPs meter data so that they can offer plans with different average (and peak) utilisation.

  3. The whole point of shaping was to stop Bill Shock.
    Remember the infamous old ADSL $29.95 200MB plans with excess data charged at $150/GB.

    Shaping is not meant to provide you with unlimited data. If you want unlimited data then chose an unlimited plan – yes they will become available, but you will have to pay for it. Unlimited at 12Mbps = 3.8TB per month.

  4. 1. Shaping is meant to hurt, otherwise it’s not a deterrent to managing usage.
    2. Why are lower speeds exempt?
    3. Because the industry pays costs on a monthly basis, so that is how its charges are structured.
    4. Why shoud some people subsidise the usage of others?
    5. Unworkable. ISPs have to pay serious money to provision backhaul, they need a business model to do so.

    This is a very egalitarian socialist approach which simply doesn’t work in the real world.

    • This.

      10Mbps shaping is just…crazy. You can pull 3TB per month with that.

      So, an ISP would have to price their 20GB plans as 3020GB plans, or heavy users would just pick the cheapest plan and max it for the whole month…

      Shaping exists so that users stick to the limit they paid for. If they don’t then how on earth can you manage bandwidth?

  5. You can propose as many ideas as you want, but as long as the CVC model exists, then the shaping model is going to be used, because the whole point of the shaping is basically to severely reduce contention, which is what saves $$$$

    Of course some ISP’s have been doing this regardless, but the CVC is just going to enforce this type of limiting. Whether or not the physical medium is fiber or copper, it has no relevance. Plans are a reflection of the pricing that ISP’s are forced to pay

    • In this case it is not just NBNCo CVC prices, but also the backhaul that the RSP needs to provide.

  6. Internode offered “flatrate” plans on ADSL in December 2002 ( and stopped the service in 2005 ( As I understand it flatrate wasn’t able to attract sufficient low and medium users to balance the leechers.

    Rather than focusing on quota, people who want higher quota plans should be focusing on higher speeds, because that will lead to people downloading more and as more is downloaded on the NBN, prices will fall (page 103 of NBNCo Corporate Plan).

    • iiNet also offered sort of unlimited plans during the time of 256k/512k/1.5mbit era, it was based on the network load, and was suppose to hurt the heaviest users, but in reality it hurt everyone.

      The shaping model is the best one, shaping is suppose to let the users know that you’re using too much, and as others have suggested it was also to be a deterrent and to stop bill shocks.

  7. I wouldn’t be too worried about how much data everyone will be getting. If Labor don’t win the next election, then most people won’t get NBN.
    I think people should be more concerned with persuading their friends and neighbors to vote labor if they want any chance of an NBN connection after the next election!

    • When I try that, I get responses such as “sure… cause fast internet should be the top of everyone’s list when deciding who should run our country….”. The problem is that few people realise the importance of the NBN for the nation’s future.

  8. Data from the same POI (local) should not be counted towards monthly usage, as its not using the backhaul

      • I reckon an in depth article on the PoI design for the NBN would be valuable.

        There are network resilience issues that have business implications etc as well as issues like unmetered data within the PoI.

        What about the location for nodes in a content distribution network?

        I think there is way too much focus on the retail side of the NBN at the moment, the network design and terms of agreements etc are far more important at the moment.

        Any limitations there will affect the way the retail landscape evolves because it will limit flexibility for ISPs to innovate and evolve their offerings.

      • Are you just being a troll for the sake of it? The posters make some good points here. You are just blowing hot air.

        • I am not being a troll. Data between customers of the same PoI is not free. Even if they are of the same ISP, if the ISP has a router on site, it still needs to pass through a CVC, which is not free. It is more likely that it does indeed need to be hauled back to their nearest PoP, so it *does* consume backhaul *and* CVC.

  9. I don’t think the tail circuit speed has a material impact on the shaping speed.

    Quotas are a proxy for the ISP to predict the network capacity they need to buy/provision for the next month. This is the reason that extra data blocks are much more expensive than regular quota, if customers can just buy extra when they exceed their quota it doesn’t help the ISP predict network capacity requirements.

    This is why something like Internode’s previous “Flatrate” plans might be interesting. Manage the bandwidth pool so customers are allowed to max their connections when overall network utilisation is low. But during high utilisation periods the top N% of customers are rate limited based on the previous few days usage.

    If you have been downloading a lot in the recent few days you are rate limited more than those that only downloaded a less.

    • THIS… until you wanted to bring back Internode’s Flatrate. Anything that complex is both hard for the ISP to implement and hard for the customers to understand. Neither will get a return on their investment in trying to work with something that complicated.

      ISP billing always reverts back to simplicity after each attempt to “improve” it.

      Capped plans with single speed shaping are a win-win: the ISP gets the predictability Cameron talks about above, the infrastructure to implement is relatively straight forward and their advertising needs very little fine print; the customer gets predictability for their monthly bills, still has basic Internet access even if they blow through their cap and worst case is they buy a data block to get something urgent done.

      Caps won’t increase with the introduction of the NBN, but as international bandwidth and transit to the “gang of four” gets cheaper they will.

      • I reckon you have hit on the very problem with Flatrate Martin, it is complex and less than predictable for users.

        It is an alternative to “traditional” shaping, I am interested in how the NBN might change some of the curent norms.

        I think predictability is the crucial element, for the ISP and for the customer.

  10. I have a suggestion. Create a billing model that bills you in steps. Only use 50gb? Get billed for the 50gb for that month. Use 100gb the next month, get billed the 100gb cost for that month. Use 500 gb, get billed for it for that month. Go away and only use 10 gb, then the bill drops down to that plan for that month. Have an option so you can cap your monthly spend if you don’t want to spend the money for a terabyte plan and would prefer to be shaped.

    Or, you can have an option to pay a bit more or less for a higher or lower shaped speed. Just to punish those kiddies for all that downloading.

    But one thing I would really like is to have unmetered data within the ISP. I’m more likely to get onto the ISP my mate has if the data between us is free :)


  11. i got to the bit where you suggested 10Mbps shaping and i stopped reading this garbage. If you want something, you gotta pay for it. Not expect more than what you paid for, for free.

    • If you’re paying for a 100Mbps pipe every month, it doesn’t seem unreasonable that it be shaped to a speed more than 128kbps if you exceed your quota. What speed would you consider reasonable?

      • If you’re getting shaped, you’re on the wrong plan. I haven’t been shaped in over a year.

      • If that is your thinking, then why not leave your access speed at 100Mb/s, but provide NO data across it, as you have used all of that.

      • But it’s not a 100Mbps plan, it’s a 1TB@100Mbps plan. You’re paying for 1TB of data delivered at that speed. Once you’ve used your quota everything else should be considered a bonus, not some entitlement.

        I’m all for an improvement in the price of data blocks, and for some plans over 1TB, but most of your shaping proposals are pretty silly.

  12. Tiered shaping, as outline by HC sounds like a reasonable proposition.
    And it has to be said, sorry Renai, article title “Five alternatives….” but the picture has SIX arrows….or am I missing something at this late hour? :-P

  13. Internode needs to release plans similar to the old flat rate plans, I’d imagine they might be quite effective for the NBN.

  14. Skype tends to be a data hog, due to its intrinsic bandwidth sharing with other users nature, and it is hard to limit the amount of data it uses unless one uses commercial data limiter applications in conjunction with it. Hence, if on very high speed plan, it may use a super lot of data, very quickly. Might be useful to have option to throttle back progressively, so that your speed is halved at 50% of total quota used, and quartered at 75% of total quota used, etc. This option would not be mandatory to take up, but would be really useful for those who are using data hog applications, which do not need all the band width one potentially have to be fully effective.

    • Stop letting it share your bandwidth with others then. There is an option to turn this off.

      • I am aware that one can limit data useage in Skype and other programs using Netlimiter 2 or similar software. I am not familiar with any option within Skype to stop sharing your bandwidth with others, though. If you describe how one would do that, it would be much appreciated, though.

        I continue to feel the option I have decribed would be useful to have an an option – though I concede not all would wish to use that option.

  15. To be fair to the ISP’s, I think they need to have customers on an appropriate plan for their usage. So I dont have issue with 128kbps shaping BUT I do like to see the ability to pay a fee to be released from the shaping e.g. optionally purchasing a data block to extend your quota until your month cycles over. If you are repeatedly finding yourself shaped, then go up a plan. Of course the ISP’s need a range of plans at different price points that makes this possible. OR, given that there’s an average cost per GB and a base level cost to supply the connection, why not offer to bill on actual usage to the nearest GB? That would be base level cost + a calculation on GB used. Perhaps also allow the customer to sets a GB ‘cap’ which they specify in the member services area so that they can specify the maximum they want to pay, with shaping kicking in if they reach it.

  16. 10Mbps shaped speed?

    So the solution is for the RSP to provide you with *free* $200+ worth of connection.

    I’m certainly interested to see how RSP’s will fight it out, but you do have to spend a couple of minutes thinking before posting every suggestion what comes to mind.

  17. 6) Pay attention to your limit and don’t go over.

    7) Pay through the nose for an unlimited plan. They will exist eventually.

    8) Shaping that slows your connection dynamically throughout the month to match your average rate of consumption with the cap you have remaining. Ensuring that you can never run out and shouldn’t ever be shaped to pathetic speeds.

  18. There is no doubt in my mind that with a change of technology pricing models need to change as well. I have an idea, But forgive me, typing and an iphone don’t go well to me. Here goes;

    New month rolls over, all customers are place in some kind of traffic class that they can max their connections with regards to pOrt speed and available ISP bandwidth. The month continues, customers that exceed their quota are placed into a sub-class where they can still access their network but only using in unutilized Bandwdth. This way, they can use the bandwidth other users aren’t but are still penalized. Users stir in their quota limit get QoS type of priority, and if all users within quota are using all bAndwidth, the users without quota are then throttled to the 64/128/256kbps speed


  19. >>2. Only shape high bandwidth services
    This has been a source of a lot of negative PR overseas because customers weren’t made aware of what services would be shaped, or other services were unintentionally affected because they used the same protocols. It could also be easily circumvented by a VPN. Not to mention the fact that this raises net neutrality and privacy issues (since it could only be implemented by looking at the packets being sent)…

    >>4. Focus on the heavy users
    That sounds a lot like extortion – ‘pay us extra or you won’t get your full speed’. People will also be up in arms about not getting what they were promised, since you can bet that the ‘top 5% of users will be shaped’ clause will be in the fine print.

    The best options to me seem to be higher shaping speeds and usage based billing. Shaping is popular for a reason – no added fees, and it works in a predictable and simple manner. Users know exactly when they’ll be shaped and what they’ll be shaped. I would guess that the speeds will go up once the NBN is actually deployed and competition starts to kick in.

    UBB is good for the same reasons, but is really only suitable for low bandwidth users when you’re charging $1/GB. It also has the potential to result in unexpectedly large bills if little Johnny starts torrenting some TV series, which is one of the reasons shaping is so popular – no added fees.

  20. If you want radically new models, you need to re-design the network (as I mentioned in another article).

    Drop the wholesale layer-2 model, it won’t work. Design the whole network as layer 3 network.

    Re-design OLTs to treat customer on the same PON as if they are on a LAN – ie, put them on the same subnet, allow the OLT to act like a switch/hub in a home or business LAN. Plug each PON “switch” into a router that routes between physically-adjacent PONs, and plug those routers into a central router (or routers) at each POI. Build a national network to connect all the POIs + peer at all existing data centres/peering sites + international links. At that point, your network is now nothing like the current wholesale networks at all – you have effectively brought the full layer 3 internet right to every home.

    Billing based on usage is now effectively impossible, but between neighbours you can send unlimited amounts of data 24/7 without impacting on anyone outside the PON. Between neighbouring PONs, you’re only loading the two PONs and the router. P2P application will now have insane performance, as long as they prioritise the local neighbourhood (easy enough with IPV6).

    The problem with such a design is that all current ISPs would be made irrelevant, and you can forget the idea of “competition” entirely. Such a network would work wonderfully under some form of socialism.

    • Drop the wholesale layer-2 model, it won’t work. Design the whole network as layer 3 network.

      No, the problem with that sort of network is that is wouldn’t be feasible in the telco type world.

      You’re trying to design a telecommunications network the same way you would a LAN, it just doesn’t work that way.

        • To clarify, it is *technically* possible, it just wouldn’t be done because it would completely destroy the concept of an ISP or a last mile network or charging for bandwidth.

          • At a telco level, no, it’s not.

            You need to remember the NBN is not just for internet connections, it must also be able to provide services capable of doing physical layer emulation, voice (akin to Telstra’s OnRamp and Optus’s Multiline products), etc.

            The NBN providing layer3 connectivity at the POI would not allow for this.

          • I’m not talking about the NBN (in its current form) here – I’m talking about a hypothetical national layer 3 FTTH network. You are trying to fit what I am describing into the existing model of how things work.

            The concept of RSPs would be gone. It would be like a national road network – you can use it to deliver whatever you want, the data never goes further than it needs to. Services (such as voice) can be provided over it.

            What services are you thinking of that can only be provided over layer 2 or “physical layer emulation” and not layer 3?

          • I am talking oddly enough about physical layer emulation services.

            As stated, the NBN is not just for internet connections. Internet type connections will be a single product available over it, there are a lot of different business type connections that also need to be able to be handled.

            You’re also ignoring that some customers for security reasons won’t want their services to go up to layer3 on the transmission path.

            And this has got nothing to do with whether RSPs exist or not, it’s got to do with the fact the services that need to be provided won’t be able to be provided if it’s a layer3 backbone.

    • Ignoring the flaws… what benefit do you foresee this giving? Fast P2P? Is P2P going to be using such vast quantities of data that the current plans are insufficient?

      • I’m not saying this should be done, I’m saying this would be a way to achieve the paradigm-shifting proposed by Renai in other articles.

        The benefit would be that any limitations of aggregation (such as contention) would be almost completely gone. A P2P model could distribute every kind of content imaginable. 128-way video conferencing? No problem.

        Where the P2P model would be unsuitable, one would be able to set up a server in every neighbourhood with “free” bandwidth.

  21. 1. On a 100Mbit plan from Internode or iiNet with 200-300GB of quota, I can use all my quota qite happily in the first day.
    If you shape we to 10Mbit I can then use a further 86.4GB per day (assuming 10 bits to the byte to cover transmission overhead), which adds up to 2.5TB over the remaining 29 days of the month.
    For that month I’ve monopolised 10Mbit of CVC, which we know has a fixed per megabit, per month cost. I’ve downloaded 2.7TB of data, taking up bandwidth of the backhaul links from my NBN POP and on my ISPs international links.
    A fair portion of my traffic probably came from overseas sources that my ISP doesn’t have bilateral peering arrangements with, so my ISP is using paid transit links to get that data for me, which does have a per-MByte cost. And my ISP should just suck that up and still provide that plan to me for less than they charge the guy on the 1TB plan? And still somehow make a profit?
    As another commenter said above, you’re forgetting that the NBN does not equal the internet. There are still significant costs to getting data here from Albania if I want to spend all my days downloading Albanian TV shows on my 10Mbit shaped connection. Someone has to cover those costs and it seems strange to argue that it shouldn’t be me.

    2. Lovely idea, how do you do it? Do you simply prioritise certain TCP or UDP ports when your backhaul links start to fill up? How long does it take before your userbase discover that setting their bittorrent client to use port 80 means their downloads go faster? Ask Comindico. There’s a long history of that not working in Australia and around the world. Alternately, you can invest in big, expensive layer 7 filtering boxes that are able to do deep packet inspection on the fly to classify traffic regardless of what port it’s on. The problem is that these boxes have capacity limitations driven by the speed of the processors and ASICs in the boxes. A box that can do DPI at 500Mbit/sec costs quite a bit more than one that can do it at 100Mbit/sec and so on. What you’re doing by putting L7 filtering devices on your peering and transit links is adding a significant amount to the cost-per-megabit of those links. That means as the ISPs grow and their traffic volume increases, it’s more and more expensive to upgrade their links out to the rest of the world, because there are more boxes that need upgrading every time they do. Various ISPs have tried both approaches in the past.

    3. This one gets discussed in depth every couple of months on Whirlpool. Because there are quite significant per-month costs in just maintaining the connection in your house, ISPs can’t use the same simple model as the pre-paid SIM in your iPad. It’d have to be something like $40/mth gets you a 25Mbit connection that can’t download anything and you buy data blocks as you need data. Ask you’ve pointed out, Internode and other ISPs are already doing data blocks, and when you combine their lowest quota plans with the ability to buy datablocks you’re partway there. The problem is that, as you say, the cost of datablocks needs to come down significantly. Is it economically viable to do so? The fact that it’s a very competitive marketplace suggests that if it is, one ISP will do it and the rest will follow suit. What datablock pricing won’t do of course is match included-in-plan pricing for a gigabyte of data. Using your Telstra pre-paid example, 10GB of data in a chunk with a 365 day expiry costs $150, while a plan with 8GB can be as low as $40/mth if you bundle or $60/mth with no bundling.
    ISPs (fixed or mobile) are always going to charge less for gigabytes that you’ll use within a set (short) period and probably won’t use all of than they will for gigabytes that you might use any time and probably *will* use all of. There are different costs associated with the two. Most people prefer to pay the lower cost per GB, so those plans remain more popular with ISPs.

    4. This is exactly how Internode’s Flatrate plans on ADSL used to work. Pool of bandwidth available to each region, and each users access to that pool was based on their personal 7-day download history. If I’d been leeching like a mofo for the last week and lots of other people on Flatrate in Canberra wanted bandwidth, my connection slowed to a trickle.
    It was an interesting, innovative plan and a great way of preventing billshock and the excess usage charges that were so common in those days. Unfortunately it also caused widespread confusion amongst customers and was eventually withdrawn when it stopped being economically viable. You know what killed it? Telstra released a 256/64k plan with 500MB and Shaping rather than excess charges for $29.95/mth.
    The problem with any of these flatrate, unlimited or uncapped plans is that you need 10 low use customers on them for every high-use customer. If you average out the costs and get a middle-of the road price that everyone pays, then it can work. Where it fails is when your low-use customers notice that other ISPs are selling low-use plans that will cover their needs and cost them less and they all churn. You’re left with a bunch of high-use users paying you less than they’re costing you an bitching loudly that everything’s so slow and they’re not getting the Unlimited they paid for. It’s something that gets tried every year or two and ends up getting pulled. TPG did unlimited 1.5Mbit ADSL1 for $50 about 10 years back and ended up pulling it. For a few years there they had no consumer-priced Unlimited plans, now they’re taking another stab at it. Go look up the history of the various Comindico resellers a decade back, or Wild IT and their equivalents a few years ago.

    5. It seems like you’re making the mistake of thinking the NBN equals the internet again here. The NBN is just what connects the customer to the ISP. The ISP then has a whole other side of their network with totally separate costs, that connects the ISP to the rest of the internet. Cost of connecting the customer to the NBN does not map directly to the cost to provide the internet to that customer. When you say Exetel’s $1/GB option was too expensive and would need to come down, how do you think Exetel should arrange that? They’re not a company known for fat margins or super profits, so you’ve got to assume that was a price close to what they thought it would cost them to provide internet services on that model. If even they’ve backed away from it, you know it’s not likely to be viable.
    Beyond that, customers like the comfort of knowing what their bill will be each month and knowing it won’t be $200 just because little Johnny has been locked in his bedroom downloading game demo and “linux distros” all through the school holidays. Look at the rapid rise of “cap” plans in the mobile market and the broad acceptance of plans with a fixed quota and shaping as opposed to plans that charge extra for extra usage. A 100Mbit connection can rack up a *lot* of downloads if you let it run. Even something as simple as a broken PS3 update that keeps trying to download the same thing repeatedly (happened about a year back) can rack up terabytes of pay-as-you-go usage in just a day or two. Cue bill shock, outrage and ACA/Today Tonight stories about ripoffs. Most consumers vote for fixed plans because they like knowing that’s the most they’ll pay per month.

    [chunk deleted because of impoliteness — Renai]

    • Hey Ryan,

      some great insights here. I did, however, delete a chunk at the end because it was offensive. I’ll let you off this time, as I thought your comment had some great content, but next time I’ll just delete the whole comment. Please see our comments policy here:

      In short, commenters must be polite.


      • Censorship sucks and breeds weakness. I’d like to read Ryan’s impoliteness, coward!

  22. I agree with Mark Newton’s comments on twitter, there is nothing wrong with the current situation with capped plans, if users want more usage they just go to a higher plan, it’s not that complicated.

    There seems to be an impression that people will suddenly go through a hell of a lot more usage simply because their speeds are faster, although usages will no doubt go up I just can’t see people who currently usage say 100GB/month suddenly deciding the need to use 1TB or more/month.

    The only change I would suggest would be giving people the option to buy some extra usage if they need it. But that I mean some people may be happy with a 50GB plan for instance, then something happens, visitors come, whatever, and they are heading towards using it, they should be able to buy a quick top up for that month without the need to go to a whole new plan.

  23. Renai,
    Most of your suggestions do not take into account the costs to ISP’s. If you want to download more, then you have to pay for it, because bandwidth costs money. That price will go down, but not just yet.

    Is your problem that you can’t get more than 1TB quota?. Then that is what your article should be about. If not, then upgrade your plan to a quota AT or ABOVE what you plan to use – If people are shaped, it means they have used their quota. The reason it is shaped to a “just usable speed” is because bandwidth costs money.

    Bandwidth costs money. Bandwidth costs money. None of your suggestions that involve giving it away for free will work because of this.

  24. Perhaps a pricing scheme similar to that of electricity companies could work, where there’s a flat service fee (including a small amount of data) and a per gigabyte charge.

    ie: a low end plan with a $30 access fee (including 5GB) and $1/GB data charge (maybe limited to $99 a month)
    a medium plan with a $40 access fee (including 5GB) and $0.25/GB data (limited to, say $199)
    and a higher end plan with a $60 access fee (including 25GB) and $0.10/GB data
    and maybe an all-you can eat unlimited plan for $200 or so.

    We don’t see fixed-price plans for other utilities (which is what Internet access essentially is), why should ISPs be any different?

    • The only problem with this idea is that it falls back to the problem where too many people don’t know, and don’t understand what a GB actually is.

      It’s the same problem mobiles have fallen into with their plans, people can read what it says but don’t understand what it means and is just going to end in more “bill shock”.

      • There’s also the fact that generally people’s electricity/water accounts are about the same every quarter so people can budget for it, whereas internet usage can fluctuate quite dramatically from month to month.

  25. what always seems to get missed is the fact that almost every customer over caters for themselves. i for instance am on a 1tb plan but i dont often hit the 800gb mark, so my slight gripe is that i pay for iinet to get 200gigs of “free” data which im almost certain will be sold to someone else as well. this would be becuase isp recognise that most ppl wont use their full quota but then go ablsolutely nuts when you eventually do.

  26. Unfortunately, there will always be those who cannot afford to upgrade to a plan with an increased limit, assuming one exists. My partner and I cap our 150GB ADSL2+ once every three or four months (usually because of guests) but we cannot afford to upgrade to the 500GB plan and our ISP doesn’t allow the purchase of datachunks. And even if we were not on a contract for another few months, none of the cheaper companies are able to service our area, or we would have chosen one of those when we moved in.

    • If I’ve already paid for it, I want to download it. Simple.

      How much of the traffic that apparently costs so much to pull from overseas is done by people that want to use the remaining 10gig on their 15gig plan, purely because they have paid for it and not used it that month. {You havent used all the petrol in your car this week, but lets charge you a full tank’s cost when you “recharge” it, who likes that idea?}

      Therefore a cap based system requires greater resources than a usage based system. But this unused download is likely where all the profit is, if theres truly a per mb cost.

      • I’m not entirely seeing your point… There’s nothing stopping you using the rest of your plan, you don’t get shaped until you hit the cap….

        The problem with a usage-based system is predicting what the maximum network load is (same with unlimited plans).

  27. One of the arguments for building the NBN was to eliminate shaping. What happened to that?

  28. What’s all this non-sense about shaping when we have TPG UNLIMITED??? Have they announced that TPG Unlimited won’t be available on the NBN? Otherwise this whole conversation is completely mute IMHO.

    • TPG haven’t announced anything, but I strongly suspect that they will continue to offer some form of unlimited (at least for 12/1).

  29. My feelings on why fixed internet access was set up with set quotas and shaping is that there was alot of grief over people who had exceeded their usage, racked up huge bills when they were not speed limited, and complained about it (just like the way mobiles have gone with their ‘caps’). Even despite the caps, person at my work didn’t realise that listening to an online radio station used up their download quota, ended up with a $500 bill for a month for excess data usage.

    I have been on a 100gb plan for over a year, and maybe once have gone over 50 gig in that whole time. I really can’t understand what people (apart from business/home business) would be needing excessive data allowances for that is so essential to life that if they get their speed shaped to something they have previously lived with (seriously, dial up wasn’t THAT long ago) that they will die without being able to access at high speeds, and expect to get it for free.

    People saying they cannot afford to increase their plan, then maybe you need to budget your internet usage like you budget your bills.

  30. The basic issue I see (and this isn’t NBN specific) is that the “last mile & backhaul” component (i.e. the bit between the ISP and the end user) costs money.

    Using the NBN as an example, that’s currently $20/Mbps.

    Taking your 10Mbps ‘shape speed’, that means a user going flat out whilst shaped is using $200 per month worth of backhaul capacity.

    Assuming the iiNet 500GB/500GB plan of $99.95, it’s clear just based on backhaul alone that this isn’t viable.

    Assuming Internet transit bandwidth costs at least as much as the backhaul component, that number becomes $400+ per month for a 10Mbps shaped service.

    These are real costs to the ISP – can anybody suggest how this would be done in a way that doesn’t bankrupt the ISP? Naturally, upping the price of the plan itself to something (maybe $499.95 per month?) that covers the costs is an option, but I assume that won’t be liked either.

    Something about having one’s cake and eating it too comes to mind…

  31. they make too much money from frustrated users ‘trying’ to use a shaped service. shaping is designed to give the user just out of reach service to entice them over the line and purchase a data block.

Comments are closed.