Managed traffic practices, excepting the prioritization of real-time traffic like VoIP, lend themselves too easily to abuse by incumbent telcos and cable distributors, who would like nothing better than to head off the impending threat of Internet Protocol Television (IPTV) that looms on the horizon, threatening their content-delivery business.

In a perfect world we could trust 'managed' traffic shaping, but in reality, and with the current business climate, it's simply ripe for anti-competitive abuse.

[updated 2009-03-31 12:06]

We have reached the point where the ISPs cannot count on a large percentage of households paying $40-$50 a month for the gigabyte or so of data that it takes to send funny emails to the cousins allowing them to ridiculously over-subscribe a line.

By not continuing to build capacity they have allowed congestion to become an issue. The backbone, and infrastructure connections are fine, it's the final leg to individual users that is congested.

The cost of network management is incredibly high, if they didn't see it as potential source of revenue then increasing capacity and staying with a traditional "dumb pipe" approach would be obvious.

I find the arguments made by large ISPs aren't very persuasive when it comes to the impact of congestion and high bandwidth users on their network. They hide all useful details of their network from the public so we're not really sure how bad the problem is, if there's a problem at all, and most importantly, what the real motivations for these management practices are.

I say, draft some strong rules to address network neutrality issues, shore up the rights of customers with respect to contracts, fees, and equipment mobility and ensure that ISPs leasing lines from the larger ones are free to compete. Then just leave it up to the ISPs to figure out how to be compliant. If the CRTC can enforce those rules and customers are free to move to an ISP that meets their needs then we don't need to focus on the specifics too much.

[updated 2009-03-31 13:26]

I ask you: Would you build a 4 lane highway with a speedlimit of 80Km/h or a 2 lane highway and slow traffic to 50km/h during peak times to deal with the traffic volume? Future traffic would suggest the first option is smarter.

[updated 2009-03-31 13:31]

Rogers and Bell have a monopoly and therefore do not feel the need to invest in infrastructure. An internet service provider should provide the service, not manage it or shape it. I pay a premium for my service and occasionally use Bittorrent, for perfectly legitimate usage. The speed is comparable to dial up. If the CRTC is going to allow monopolies, as it clearly does, it should insist that the ISP keeps up with technology and demand, allowing fair usage for all. They have no business managing traffic. Neither does the CRTC, whom I blame for this situation. Simple.

[updated 2009-03-31 13:32]

It seems to me it's almost a right that we should get unaltered access, or at least what we paid for. It is inherently wrong to control the spectrum and gives too much control to greedy companies.

[updated 2009-03-31 13:40]

If forced to choose from the options presented the best approach would be the Best Effort model coupled with Network-wide management only when necessary.

However, after having studied this issue in depth for the past few months and also being directly effected by it, I think that the best solution would involve something different than the proposed solutions.

ISP's have to change their approach to dealing with high-bandwidth internet use. Severely limiting the ways in which the user can use the connection that they've purchased does not solve the problem because technology will always evolve, new protocols will be developed and workarounds discovered. New high-bandwidth uses for the internet are going to be created and we can't just ban their use so that the telecom industry doesn't have to invest in new infrastructure.

Rather than trying to make the end-user's behavior fit their service, they should change their service to fit the end-user's behavior. In a truly competitive environment this is how things would function, and ISPs would see this congestion issue (if it even exists) as an opportunity to innovate and beat the competition, rather than a problem to be solved by crippling their own product.

[updated 2009-03-31 13:44]

I strongly support what you describe as "best effort to all", a protocol-independent approach subject only to overall capacity / throughput limitations purchased by the consumer. The "managed" approach leads to inevitable qualitative decisions of what is "best" - which equally inevitably becomes "best for the supplier" not "best for the consumer". When the capacity that is demanded - which at that point has in fact already been sold - exceeds the network capacity, then the revenue from those sales needs to be directed into upgrading that backbone to be able to continue to serve that capacity and the community. Otherwise, in the "managed" scenario, throughput is throttled "in the interests of all" and the excess revenue is channeled into corporate profit. In the case where these major suppliers are virtually monopolies, the role must exist for government acting as the agent of the population to assure fairness - which does NOT translate to the ISPs implementing "management" policies that turn into marketing tools.

[updated 2009-03-31 13:49]

A total ban on ISP's conducting any throttling or shaping based on packet sniffing is the only plan forward. (just like the ban the FCC placed on Comcast)

It's up to ISP's/Telcos to provide the necessary bandwidth and network infrastructure to support use of the service, not to deploy a service rated at a set speed and set download limits, and then further restrict the type of traffic customers use.

This is a clear violation of consumer privacy laws, and places the ISP's in an extremely dangerous position of having the discretion to determine what type of internet traffic they do and don't have to provide a standard quality of service for.

[updated 2009-03-31 13:52]

Use the "Best Effort for All" approach. If costs have to increase, and passed along to the consumer, market forces will adapt the services provided to the needed requirements.

[updated 2009-03-31 13:57]

The Best Effort to All Approach is the only approach that is fair and results in a competitive market for Internet connectivity. Regular phone companies don't change your phone service based on the type of speech you use, so why should ISPs use throttling on any type of use of their service, especially if they are offering certain bandwidths for certain prices?

[updated 2009-03-31 14:24]

i think a management approach which gives fairness between users, and priority to interactive applications over non-interactive applications is best.

isp's should not manage content, but should manage the delivery vehicle to ensure the best experience possible to the largest number of users within the constraints of capital and capacity.

i don't agree with the definition of 'over-provision' above. that would require a 1:1 provisioning, e.g. 1Mbps to the home would be 1Mbps all to the core. This will give a very poor experience to the user. I would rather have a 10Mbps burstable connection than a 1Mbps guaranteed.

I do not believe that usage charges are an effective means of controlling capacity. Capacity is a peak, usage is an average. Giving me a limit of a number of km / year i can drive does not limit the amount of rush hour traffic i cause, it just limits the off-peak trips i make that would be optional.

[updated 2009-03-31 14:59]

Application-level management should only by applied in the case of giving technologies that need higher priority in order to function correctly the necessary priority. Examples are VOIP and video conferencing. All other applications should be treated equally, without any merit judgements based on what a client is doing. Network-wide management should be the norm for most applications.

Subscriber-level management should be used to force ISPs to provide the capacity they have promised to their subscribers and to limit subscribers to their purchased capacity. The fact that the CRTC permits false advertising of ISP bandwidth is criminal. (The famous "up to" clause).

Building a network that can handle the demand generated is a sound business practice, and a successful ISP business will move to fill this void as a business priority, or lose its business to another provider. Already ADSL providers are losing out because they did not want to invest in infrastructure and their service capacities have been equalled or bypassed by other types of providers.

[updated 2009-03-31 15:01]

The "best effort to all" approach is the only acceptable approach.

The ISP's may have a legitimate complaint that traffic requirements are increasing too fast to keep adding infrastructure however they should then re-evaluate their business model to ensure profitability and ongoing infrastructure improvement. Controlling the type of traffic should have nothing to do with this. Maybe ISP's should stop marketing "unlimited" bandwidth for $40/month and charge a premium service for increased bandwidth. This would ensure everyone has access to a minimum acceptable level of bandwidth and so called "traffic hogs" can pay for all the bandwidth they require.

It is not up to the ISP to decide which type of traffic is more important, this is contrary to the design and expectations of the internet. What if one ISP arbitrarily throttles access to traffic directed at one particular companies services because they are a competitor? It would be like building road blocks and toll booths around all the streets surrounding a competitor’s office building.

It is also extremely important to consider where this may be heading. "Traffic shaping" could possibly lead to an artificially created pricing model where companies could purchase the right to elevated access for their traffic garnering them an advantage over smaller rivals. This may sound great from a business standpoint as it creates an artificial market and new revenue stream however it would be extremely unethical and should not be allowed.

[updated 2009-03-31 15:03]

As P2P, video streaming and other network traffic demanding applications gain in popularity I think its time to overhaul the way we view the internet and its traffic. As a heavy P2P and video streaming user I view being on a high-speed network a priviledge not a right and therefore would like to see the rates paid reflect that. Here's what I propose

Have a two teir system. The first (for low demand users) a set monthly rate depending on the speed subscribed too with a per GB rate charge if gone over a set capacity. The second no monthly rate (or a base minimal rate the same across the board) only a per GB rate that slightly increases as higher speeds are subscribed too with an option to buy chunks of capacity at a discount (the smallest chuck would be bigger than the biggest capacity allowed for low demand users)
with no sort of throttling.

I would also like to see some sort of off-peak/peak rates set for per GB rates

treating internet use as you would electricity use i think the best and most fair option out there, lets everyone surf the net at the speed they are paying for (or very nearly) and lets users control how much and when they use the net.

[updated 2009-03-31 15:05]

I do not think BELL has proven that network is congested. No one outside of BELL has privy to their service roadmap for the future, but it is not hard figure out that BELL/Roger's 1st step to ensure they not taken out of the loop by ensuring they don't allow to much bandwidth downstream that could let other more innovative providers usurp of traditional businesses (like telephone, PayPerView, TV) by providing same or better service via internet.

The biggest deception is having the general public think the networks are overloaded. Ask ANY 3rd party ISP if they have the ability to buy more "bandwidth" from InterNAP or what have you.....

Again, the conflict or interest is clear here in regards to BELL and ROGERS/SHAW

[updated 2009-03-31 15:08]

I think all of the traffic management techniques mentioned may have their merits except "service or application level management". This is the method currently being used by Bell (although they are also proposing a form of "subsciber level managemet" through the UBB tariff they submitted completely ignoring of CRTC Order 2009-111). I object to service level management because it is against the principle that internet should be provided in a neutral manner. Not everyone will agree on what applications should or should not be given priority so some users will be left unhappy. A more fair approach is a type of subscriber level management I will discuss below.

When I lived in the UK I used Virgin Media cable internet. There they employed a subscriber level management that would cut your speed to about 1/3 your subscribed level during peak hours after you had downloaded 2GB that day. I felt this was a fair management practice that was neutral to users and appropriate to the shared nature of local nodes on cable internet. THIS IS NOT VALID FOR DSL INTERNET WHERE YOU HAVE A DIRECT UNSHARED CONNECTION TO THE CO!

Aside from my philosophocal and moral objections to service or application level management due to its inherent non-neutral nature, I believe ISPs should be able to handle their network the way they see fit. Key to this however is that there should be sufficient competition in the market for there to be a choice for end users as to which network they wish to subscribe to and consequently which traffic management suits their needs. The current practice of allowing the last mile (GAS) provider to dictate traffic management precludes the possibility of such choice and such competition. As I mentioned above last mile management is unnecessary for DSL due to its nature as a dedicated connection to the CO.

There is obviously a lively debate as to what traffic management practices are most effective and which are necessary. The solution is to foster competition and allow ISPs to implement what they see fit so customers can choose what suits them. If a practice doesn't work and a network becomes congested, the customers will leave and the practice will disappear. What is invalid is to allow a GAS provider with obvious conflicts of interest (owning both the last mile and a retail internet service) to dictate what management practice will be used by EVERYONE and thereby stifling competition, innovation and experimentation in what practice is truly best for providers AND consumers.

[updated 2009-03-31 16:32]

Practices such as throttling internet speed where they have clearly advertised faster speeds all in the sake of saving money instead of upgrading infrastructure actually encourages people to not use their services. An ideal world for an ISP would be to have many subscribers not using their services. They wouldn't have to upgrade their equipment/lines and they would rake in their monthly fees. However these companies advertise their amazing services and people actually want to use it! The companies then say we have too many people downloading and then throttle the internet connections. The CRTC should mandate that Internet service providers at all times shall allow equal and unbiased service to any user and any application that they may be using. Thus forcing companies to upgrade the infrastructure to support these uses thus creating innovation in internet applications.

[updated 2009-03-31 18:21]

I would like to see the delivery of internet content treated just as a utility service like water and electricity. Most people pay for the amount of water or electricity that they consume in a month, why can't that approach be used for the delivery of digital data?

Traffic shaping is unavoidable. It is imperative that applications such as VOIP have priority over email, for example. A network-wide management approach where the whole thing slows down as it becomes congested is not useful because it will effectively kill some important applications such as VOIP.

A combination of subscriber-level management and limited packet prioritization is best. Having the subscriber pay for the amount of data they transfer will reduce the "abuse" that P2P creates while prioritizing packets will guarantee that important services aren't affected by possible network congestion.

[updated 2009-03-31 18:28]

Deep packet inspection is equivalent to the post office reading my letter to decide which priority to give it. The post office cannot read my mail without a court order. The same should be true for ISPs.

[updated 2009-03-31 20:59]

Comment removed at participant request

[updated 2009-03-31 22:46]

The internet is a physical network with costs roughly proportional to traffic carried, similar to a phone network. If I need more phone capacity, I pay for another line. Now while this is simplistic and ignores bandwidth expansion technologies like DSL, eventually this is the model that the Internet runs on. Bandwidth is what you pay for. So it seems fair and commercially reasonable for ISPs to charge based on usage. I would suggest that a minimum fee (e.g. $40/month) cover up to say 60GB per month and after that usage would be proportional - the next 60GB would be charged no more than another $40. To imagine any other scenario is asking some users to subsidize others - and if we want to do that, that should be done with tax rebates and decided by politicians. It has no place in a fee structure. And before anyone asks, I am a user with no connection to any ISP.

[updated 2009-04-01 08:58]

Since putting down cable is a fixed cost, is there any way we can entice telcos to just increase bandwidth capacity over time, instead of increasing management?

This model is very successful in Europe where telcos are out performing in not only speed by 3 to 4 fold, but monthly allowance.

[updated 2009-04-01 09:05]

"Subscriber-level management - for example, applying usage charges to subscribers when they use more capacity than their service plan allows or offering different tiers of Internet service to subscribers."

Many ISP's already apply this standard. It is in my view that ISP's are responsible for keeping up for demand. In 2008 all the major ISP's recorded record high profits. They should be forced to implement "Network over-provisioning" with regulation on subscriber rates.

[updated 2009-04-01 13:36]

This is a loaded question.
There should be no traffic management. All applications and protocols should have
maximum bandwith available to them.
However if a choice is to be made "The Best Effort To All" approach seems to be the
best of the two. This premise, though, assumes that networks are being upgraded and
expanded. They are not; at least not very quickly. One has to only look at Sweden and
Japan to see what network expansion and upgrade is all about.

[updated 2009-04-01 16:16]

The best thing for ISPs would be not to oversell what it can offer. There seems to be a megabit race to upgrade customers to faster and more expensive connections but seems to result only in more traffic management that slows the connection down. Being sold a 5mbit unlimited connection should be a 5mbit unlimited connection, no cap in usage and no packet shaping. I shouldn't have to hire a lawyer to understand the fine-print clauses just to know what I am really buying.

[updated 2009-04-01 17:54]

I spent about 40 years in the telecommunications industry. I believe that there is a lot to be learnt from the past. There has always been congestion in the telecommunications networks from the days of the manual switch board. The approach was always "best effort". If there was not a circuit for the call, the call was dropped. I believe that this approach should be used on the Internet. The IP protocol was designed as a best effort and should remain that way.

Packets should not be delayed on any pretext. The packets should be discarded if a forward path is not available.. A person like myself would does a little browsing to get specific information will hardly be impacted. A person who downloads a Video may have significant problems. Also a person who uses Voice over IP may have significant degradation in speech quality. The Internet was never intended for these applications.

Further the Internet is a mesh network which does not have a set path between a sending and receiving node. Rigorous steps should be taken to ensure that the IP protocol is not modified to simulate a circuit switched network. Such a modification to the IP protocol would have a substantial impact on the through put of the Internet.

I suggest that my two suggestions should be given serious consideration

Paul Williams, Ottawa

[updated 2009-04-01 19:31]

If the ISPs offered realistic pricing and packages there are varying segments of the market that will pay The problem is that there is not enough real competition and hence choices and prices for consumers are poor. Bell will charge customers for a 7mbps service and only provide them with 1.5mbps of speed. For many customers they are definitely not getting what they pay for but for some reason the ISP is allowed to get away with this, so the line that costs may be passed on to subscribers makes no sense since Bell has never had a problem with passing costs on to customers. I've seen my Bell DSL service get worse in terms of speed and service over the last 2 years and during that time I've also seen the cost of the service increase.

I think because there is no real competition in Canada the best approach is Best Effort. Simply because the monopoly cannot be trusted to provide Canadians with a quality and affordable service that is priced and packaged based on free market conditions. If a managed approach is allowed, it will not "managed" in the best interest of Canadians and because there are no other choices because of lack of competition (i.e. you couldn't switch to a best effort provider) and innovation and broadband access would suffer significantly in Canada.

[updated 2009-04-01 23:31]

my view on this is quite simple i pay my monthly fee to my ISP and i expect a service in
return . What i do with the service i have paid for is none of there business .
Would we put up with the post office opening our mail to see if and when they would deliver it , to me there is no difference in the service provided by the ISP's and the post office
I want what i pay for without any discrimination in where i go on line or with what I download. If they are having a problem with traffic congestion ,which i seriously doubt,,then it is there problem not mine . could it be that what we here is simple greed and wanting total control .
I paid for it and i want the service i paid for simple as that .

[updated 2009-04-02 02:21]

How is it that they don't view over building as a adequate solution. My mother pays bell $60/month to send email and browse the web. I am sure a lot of subscribers use under 5 GB / month of bandwidth.

The truth is this, for the past 50 years, major telecommunication companies have been used to every house paying $50/month for a land line. Then cellphones came along and they got used to collecting $50/month from everyone over fourteen. The problem is unlike the phone and cellphone market, the infrastructure needs to grow. This is throwing off their massive profit margin and they really dislike that. So They want traffic shaping.

[updated 2009-04-02 02:35]

Carriers MUST remain neutral and just carry packets from A to B and not try to guess what the packet will be used for once it arrives at destination.

What I do with a packet once it arrives at my computer is NONE OF THE CARRIER'S BUSINESS. The carrier has no business trying to guess the name of the aplication which generated or which will process the packet upon arriving at destination since those activities are outside the carrier's jurisdiction.

The job of networks is to deliver packets to destination. The Telecom Act clearly states that the carrier cannot change thge purpose of a communication, and thus the cannot decide that certain packets are to be disrupted beause the carrier decides the packet won't be needed urgently by the destination.

Next thing, Bell will start to peek into bank networks and start to disrupt transactions that send money to Bell's competitors.

More importantly: when somone buys X amount of bandwidth, the carrier MUST be made to provide the bandwidth sold. If it can't provide the advertised bandwidth, then it must be severely fined by the government and be forced to adjust its advertising accordingly so that it only advertises a service it can provide.

Once ISPs are forced to divulge the exact amount of bandwiodth they are able to deliver, then canadians can compare the services and decide which service is best.

Right now, ISPs are allowed to advertise fantastic speeds they are unable to deliver because they have not upgraded their networks to support the capacity their are advertising. In a normal world where the CRTC was not a division of Bell Canada, this would be called FALSE ADVERTISING.

The fact that they are systematically throttling during fixed periods (or all the time) means that the purpose of throttling goes beyond the management of temporary periods of congestion. This is really about the control of what people do on the internet, especially protecting their plans to deliveer entertainment thyrougfh proprietary paid services which would not be viable if users continue to be able to choose their content from anywhere on the internet and purchase content from anywhere on the internet.

The internet has lived with congestion since day 1 without the need for fancy DPI/throttling equipment. Proper network management adds capacity to match the demands and advertised services.

The current situation exists because the big ISps have been allowed to advertise speeds that they cannot provide, so they are using DPI to restrict users so that they cannot make use of the advertised speeds unless they use specific services approved by the telco. This is NOT NETWORK MANAGEMENT. IT IS SERVICE MANAGEMENT.

[updated 2009-04-02 03:05]

Perhaps the most troubling aspect of this situation is that there are no alternative options (considering most of Canada only has two legitimate ISP providers in Bell and Rogers) if you don't like the actions taken by your ISP.
I recently switched from Bell to another provider because of Bell's practice of throttling users by using deep packet searches of their network usage. However (at least in Toronto), alternative providers simply rent the lines from from either Bell or Rogers. Bell has since decided that they have the right to throttle ANYONE using their lines regardless of terms of the third party provider. What this does ultimately is limit users to a monopolistic internet landscape dictated by the whims of a few (in this case 2) providers. I have no problem with Bell wanting to do this with their own customers, but they shouldn't be allowed to dictate the terms of third party providers in free and open market. Let the free market decide who's policies are best - third party providers should be allowed to do whatever they want with the bandwidth that they have legally purchased from Bell, and customers should be allowed to make an informed choice on which service suits their needs. At present there is no option...

[updated 2009-04-02 10:50]

I think that it should be left as a Best Effort to All approach. If we go with the application management, then I can see corporations like Blockbuster (with their new online video site) paying off the ISP's to ensure that their site content gets a higher priority than one of their competitors. So I would get shafted if I wanted to use something else. No Thanks! I want my internet un-managed and unfettered thanks.

[updated 2009-04-02 11:13]

Firstly a network ought to be built to be used. Keeping use at <100% is wasted infrastructure. (who paid for this infrastructure and with what public allowances?)

Prioritization of real time data (voip, http) is nice, but must be vendor agnostic. There should be no difference between bells VOIP and my application, in terms of speed or avaliable throughput. That is to say, a telco's VOIP app should consume the same limited bandwith as mine. A telco shouldn't offer free VOIP data on a limited plan, while other VOIP apps consume data.

Prioritization should be based on protocol, and not on deep packet inspection, which is an invasion of privacy. When protocol is insufficient, no management should be performed. A hands-off approach leads to the best use of bandwidth, as stated, most protocols handle traffic management internally.
A good example is encrypted traffic. You cannot know what is being carried, and it may be important (work from home). Limiting this as bad data removes the ability to (for example) work from home.

[updated 2009-04-02 11:50]

First as a background to my comments, I have been an internet user since 1994 using a 14.4k modem that I convinces my mother that I really needed to have. As well I was a very early user of the Rogers Wave cable modem service and I eventually switched to the Bell Sympatico 1Meg modem service and have switch back and forth for various reasons. Currently I am a Bell Sympatico user who is on an older 3 Meg profile with a grand fathered unlimited account.

As has been mentioned before, Bell Sympatico has been throttling specific P2P protocols like BitTorrent for the past year or two. During this time the following behaviour has been observed by other users and myself. Around 4-5pm EST to about 2-3am EST the upload and download bandwidth is capped to about 20-30 kb/sec, outside of the capped time I have been recently been getting about 650 kb/sec download and 150 kb/sec upload. This has proved to be a significant impact on legitimate uses of the bandwidth that I pay for like:

- Downloading patches for some Massively Multiplayer games like World of Warcraft
- Downloading Linux images for installation
- Downloading videos through some legitimate video like CBC's availability of some shows through bit torrent.

It appears to me that these providers are not interested in creating products that deliver as promised, while generating the margins they want and allowing for the investment in infrastructure. Instead we are saddled with companies who over sell their network capacity and count on the flawed belief that there will be enough Grandma and Grandpa types who pay $30-45/month just to send a few emails and browse a page or two.

So if you want to know what is the best approach to traffic management? Choice. Give us options to cater our internet experience to our expected usage and if we can't afford it let us change our usage to fit our budgets. Having only 3-5 plans is hardly enough to cover the different type of internet users out there. The squeeze your light internet users approach to earnings and profit is failing and the correct solution is to properly provide choices to let the customers decide what they are willing to pay for and are properly priced.

[updated 2009-04-02 12:18]

Clearly, the optimum approach has to be over-provisioning. I don't buy the argument that capacity is over-subscribed, or even approaching it. Many of the numbers filed in the Bell case last year indicate a significant capacity still available. In the long run, the capacity will have to be provisioned, unless we are to be stuck in a http based world when everyone else in the world has moved on to a richer internet world.
As you correctly indicate, many services throttle by their very nature anyway, and ultimately, most well designed modern protocols will include mechanisms to control their flow. I would expect next generation protocols to adhere to this mechanism.
If management has to be imposed as an interim solution, it should be uniform, and non-protocol discriminatory. Usually subscriber based solutions based on the bandwidth purchased by the consumer should be amply adequate to provision such throttling were it needed.
The other question, not raised in your discussion point, is when such throttling should be applied. The major issue I see is that throttling at the moment is universal, complete and without escape. It is based on the optout principle as well (such that throttling is the default state, and certain protocols are opted out of throttling). This is unfair and unnecessarily draconian- throttling, if it is applied at all, should be short term, to deal with transient high volume loads. The implication of the current throttling regime is that the capacity of the Canadian internet is about 1% of demand for it, which is patently ridiculous and not supported by industry analysts.

[updated 2009-04-02 12:30]

I'm strongly in favour of the "Best effort to All" approach.

As others have said, the advent of VOIP, VPN, and online video have resulted in more customers using the full potential of their internet connection.

As an internet service customer, I fully expect that I should have full control over the bandwidth that I have paid for. I also expect that I should get the full allotment of bandwidth that I have paid for.

While I don't expect ISPs to have dedicated bandwidth for each connection they sell, I do expect them to scale their networks so that they can support the bandwidth used by those customers who choose to use the full potential of their connections.

[updated 2009-04-02 12:46]

The "Best Effort to All Approach" is the best way forward for broadband subscribers. Clearly the best solution is to have a network capable of handling the subscribers that an ISP has signed up - after all, the ISPs have the option to stop adding customers and terminate service to their largest bandwidth-hogging users to bring their load in line with their network's capabilities.

To keep subscription costs down, I can accept that ISPs want to incorporate reasonable traffic management, however the three methods listed in the "Managed Approach" outline are radically different. I consider two of them completely unreasonable:

Network-wide management doesn't make much sense to me. The ISPs claim that the majority of congestion is due to very few heavy users. Why should I be penalized because someone else on my cable node is one of them? Consumers have paid for a certain grade of internet service, and they should get what they have paid for.

Service or Application-level management is unacceptable for two reasons:
1) It is ripe for abuse. If ISPs are allowed to slow down specific applications there is a strong incentive to, for example, slow down all streaming video and then create an exception for their own video service. Similar "approved services" business models are already harming consumers in the mobile phone industry, where exorbitant fees per kB of data are the norm and ISPs offer to exempt certain "partner" sites from these charges. When the partner site isn't owned by the ISP outright, they are often subscription services that the customer must pay, and who then kick back to the ISP to get on their "partner" list. The ISP makes money coming and going, while the customer ends up paying far more than they should.
2) It simply won't work. The single biggest cause of network congestion is P2P apps and most P2P traffic is now encrypted, meaning ISPs can't determine what type of traffic it is in order to throttle it. Throttling all encrypted traffic is not an option, either: there is no way to tell an encrypted P2P app from a corporate VPN or someone doing online banking. If it can't stop P2P congestion, there is no point in this approach at all.

The only traffic management policy I consider somewhat reasonable is subscriber-based. Each customer signs on for a specific speed of service, with a specific monthly traffic cap (or none), and a clear statement of what happens if they go over that limit (speed degradation, extra charges, etc.). This allows ISPs to make Johnny Bittorrent pay a fair price for his bandwidth-hogging habits without requiring Grandma Email to subsidize his connection.

However, I say somewhat reasonable because I do not believe even this approach is acceptable in today's broadband market. Most people have only two choices for broadband: cable, or DSL. There is only ever one cable provider available. While line-leasing allows for third-party DSL ISPs, Bell has made it clear that all third-party traffic is subject to the same management as its own network. So there are only two companies setting traffic management policies for any region, and you can bet they're watching one another and matching policies. I would pay significantly more than my current ISP's fee for a completely unlimited, unthrottled connection to the internet. I do not have that option - all ISPs in my area either cap their customers, throttle their traffic (or are on bell's lines and bell throttles their traffic), or both.

Until customers are given true competition in the broadband market and have the option of voting for open networks with their wallets, ISPs will continue to oversubscribe their networks, overcharge their customers, and maintain that there is no demand for open internet connections without ever offering one.

[updated 2009-04-02 14:15]

"The Heavy Reading report (see Document Library on the right side of this page) on traffic management technologies suggests that the majority of ISPs do not believe that over-provisioning is an adequate long-term solution for managing ever-increasing traffic volumes. When ISPs build additional capacity, there are costs, which may be passed on to subscribers.".

It is a free market so of course additional costs may be passed on to subscribers.
Market forces should determine costs.
Maybe a majority of ISPs (I hope a statistical survey was done) do not believe over-provisioning is an adequate long-term solution but maybe over-provisioning is the best solution for consumers.

[updated 2009-04-02 14:22]

ISPs should offer best for all.

right now it is false advertising to offer speeds that are not delivered. if you pay for a service that advertises certain speeds you should get it.

if ISPs want to limit that should be upfront and honest about what they limit and the consumer should be able to make their choice of ISP based on that as well. But for that to work there needs to be more competition in the Canadian ISP market to see that consumer get a fair deal - which they are not at the moment.

Traffic shaping is an effective way to control bandwidth use - go over your 60GB allowance in a month and be speed limited to 3/4 of promised bandwidth. go over 100GB and be speed limited further to 1/2 promised speed. customers who want more can choose a plan that offers more data allowance.

[updated 2009-04-02 15:09]

they clain that network over-provisioning is not a good way to go, but we feel that it is the proper way to go. if you know that traffic on the internet shows no signs of slowing down, and more and more people will start using it, why not get ready for it now?? why wait when its already too late. the first method is by far an aceeptable method, but the ISP's will always employ the DPI technologies that invade our privacy.

[updated 2009-04-02 15:13]

The best effort to all approach makes sense. We live in the 21st century, and high speed internet access is essential for our society to thrive. Compared to the internet access speed in Asia, where > 10mb/sec access is the norm, our access speed is woefully inadequate. Allowing the big cables / telecom companies to manage our traffic while avoiding more investment is going to hurt our productivity as a society. This is especially important, considering the fact that the US Obama is investing heavily on internet infrastructure now.

[updated 2009-04-02 17:20]

I understand that bandwidth costs, but the Internet should be considered public infrastructure, and whatever the decisions are made should be made in the best interest of society, not in the best interest of ISPs. Sure they built it (to various degrees), but they need to be compensated for (some of) their previous investments and future investments should be made by the public. I'm typically loathed to hand something over to the government, but can you imagine if every road in Canada was a toll road? Basic Internet infrastructure and connectivity should not be solely in the hands of private enterprise.

[updated 2009-04-02 19:53]

I believe that the best effort for all approach is the best approach for two reasons.

1) It allows for the ISP the room to expand with its user-base.

2) It doesn't give the internet a "class" structure with those sites at the top being able to dominate the market it targets, which therefore maintains what the whole purpose of the internet originally was. To transmit and recieve data.

NOT to transmit data to my ISP, have my ISP decide what data is most important, have it throttle my connection based on what it deems to be more important, and then finally send my data.

[updated 2009-04-02 20:37]

I think Network-wide management is the best approach. ISPs are selling access to the internet and should be letting the customer determine how they want to use that access. Prioritizing one user's VOIP call at the expense of someone else's streaming video is an arbitrary decision based on assumptions that the VOIP call is more important than watching a video. However, the ISP can't possibly know the context in which these things are taking place.

I don't have a problem with packet prioritization in general, but I do have a problem with ISPs doing the prioritizing. They don't know what my priorities are and I don't like that they think they do. The obvious solution is to let the customer determine which packets get priority. If the ISP has to throttle down everyone in a certain area because the network is getting congested, that's fine. But even with that reduced speed if someone wants to set their P2P port as a higher priority than their VOIP port, they should be able to.

It's all well and good to suggest more network capacity, but obviously that can be an expensive undertaking. What I would prefer is simply more information: for instance, an advertised minimum speed in addition to the maximum that is advertised now. That way, customers would have a better idea of the service they will receive, even during peak hours.

[updated 2009-04-03 07:10]

I understand the need for bandwidth caps - you can't let one person use constantly when you are overselling your network because it impacts other customers negatively. On the other hand, bandwidth used at 4am is very different from bandwidth used at 6pm. I would like to see a system where there are separate caps for peak and off-peak bandwidth, or perhaps counting off-peak traffic at half the rate of normal traffic.

[updated 2009-04-03 07:23]

I believe that the Best Effort to all approach specifically the network over-provisioning approach is the best. Any traffic management should be conducted on the end users lan as the isp's links will not be congested.

[updated 2009-04-03 10:36]

Selon moi, l'usage du Network Over-provisioning est la meilleure maniere d'y faire: il ne devrait pas etre possible pour les compagnies de reduire la vitesse obtenue en dessous de l'offre qu'ils font, ie, si ils disent 5mbp/s, et que la connection entre le client et la compagnie supporte cette vitesse, il devrait toujours etre possible d'obtenir cette vitesse (considerant que la source internet utilisee fourni suffisamment de largeur de bande)

Une connection internet illimitee devrait vraiment etre illimitee.

Si le managed approach est utilisee, le service or application-level ne devrait pas etre permis, puisque certains usages qui utiliserait enormement les service priorise aurait un bien meilleur service que celui utilisant de nouveaux protocoles..

Le network-wide management est acceptable dnas le cas ou c'est 'temporaire', ie, jusqu'a ce qu'ils se procurent plus de largeur de bande.

[updated 2009-04-03 10:50]

ISPs should be neutral, accommodating all kinds of network traffic between all users. A "Chinese wall" should exist between the connection service and any Internet content service or competing content service (e.g., cable television competing with content delivery over the Internet).

Network management (of any kind) is not a substitute for ensuring adequate infrastructure is available to meet current and reasonably anticipated future demands. ISPs should be required to provide subscribers with regular reports of actual throughput vs advertised capacity, including actual throughput during peak usage periods. The use of Network management to control traffic implies the ISP has oversold its capacity.

Network-wide management is obviously necessary if network capacity becomes overloaded. Good planning should minimize the requirement, but it cannot be eliminated.

Service or application-level management should generally be shunned. This is especially true if the ISP provides content services that might appear to be given preferential treatment. There is a case for assured QoS for some realtime uses (streaming audio/video), however, this is a pandora's box and is open to abuse.

Subscriber-level management is acceptable for both network management and subscription management, provided there is adequate disclosure of how it is applied and its results. This management should only be in the form of temporary degradation of service. Neither disconnection nor extra charges are appropriate.

It is entirely reasonable for Internet Services to be sold with a quota on high speed access after which the effective connection speed will be reduced. However, due to the nature of the Internet, there should not be extra charges or disconnection when a quota is exceeded.

Subscribers cannot control what packets are sent to their computers (e.g., denial of service attack) nor what upload requests are made (e.g., slashdot or digg effect). Malware, misbehaving applications, and unauthorized WiFi use can reasonably be expected to use bandwidth without the subscriber's knowledge. Neither a hard limit, cutting the subscriber off from Internet services, nor extra charges for use over some limit are appropriate penalties when the subscriber may not be aware of the activity.

Aside from such uncontrolled traffic, a subscriber may reasonably have difficulty understanding their own usage. Internet usage has no easy human perceptible usage metric like the duration of a phone call. Again, notification and degraded service are the appropriate responses, not disconnection or extra charges.

In terms of network management, service should only be degraded during heavy usage, when other subscribers would benefit. This will encourage subscribers to shift heavy use to off peak periods, when possible. Degrading service during periods of low overall usage is not traffic management.

Heavy users should be identified by their use during heavy use periods. Users who confine their heavy use to off peak periods should not have light use degraded during heavy use periods.

Network management imposed by the ISP on their own subscriber is one thing. Network management imposed by the "last mile" provider on the subscribers of an other ISP is another thing all together. ISPs should not have their network management practices imposed by the "last mile" provider.

[updated 2009-04-03 17:49]

I am 100% in favour of Network over-provisioning, Reliance on standard protocols, Network-wide management, and Subscriber-level management. Service or Application-level management gives the telecommunications provider carte blanche to read the contents of private individuals' communications for their own purposes, and puts a huge damper on freedom and innovation.

[updated 2009-04-05 06:28]

Don’t Burn the Forest to Kill the Wolves: Future of Network Neutrality in Canada

Heated debates about whether Canada should have controlled or uncontrolled Internet can be analysed in the framework of means and ends. It is possible because ultimately controlling or not controlling involves certain means that are aimed to achieve certain end results. While there are strong arguments made by both the proponent of controlling, (mainly consisting of Internet Service Providers) and opponents of any kind of regulations (mostly, end-users, privacy concerned civil society and some academia), the issue has not been resolved yet.

Proponents want to control and benefit to the maximum from their investments, whereas the opponents cannot tolerate any control whatsoever over the flow of data through the Internet. In this piece of writing, I would like to discuss the possibility to find an “equilibrium” in this controversial subject. I believe this debate will not produce a result which will satisfy only one side’s interest, but rather the result has to be mutually-satisfying. This means that “the pie will be shared.” i.e. ISPs will have some control while allowing reasonable unimpaired flow of data. If such a result can be realistically assumed, then the analysis of the matter should be concentrated on what can be that mutually-satisfying result.

In that respect, I think, the following key points should serve as guidelines in adopting desired result:
Respect to existing laws and the Charter: Most arguments against Internet traffic management techniques are concerned with such issues as freedom of expression, freedom of choice, privacy, rule of law, objectivity , democracy and free market economy. Canada, as one of the most developed countries in the world, has achieved certain level of progress in these issues. This progress should never be compromised by the desires of those who want to use their techniques to control the Internet traffic.

Reading the posts on the online consultation forum organized by the Canadian Radio-television and Telecommunications Commission (CRTC), one may conclude that traffic management techniques will certainly produce some kind of discrimination. If this is the case, then one should also not forget that in Canada there is no absolute protection from discrimination. Our law allows us to discriminate, provided that an objective of discrimination is pressing and important; the means employed are rationally connected to that objective; the discrimination has minimal entrenchment on our rights and freedoms and that there are no less entrenching alternative to the discrimination. I will briefly go through these conditions.

Whether the objective of discrimination is pressing and important? – ISPs claim that because of network congestion, (often cause by ill-motivated users, or even by normal busy use of the network) the ISPs need to discriminate (or prioritize) certain sources of data and applications online. In another words, the ISPs do not want to use 100% of the capacity of the networks that they have built for free. This implies that if you pay them, you can, at least theoretically, use the whole capacity of the network. It seems that the network congestion is not that all-affecting problem that needs to be resolved. It only affects the ISPs who have built the networks with 100% capacity, but now they do not want to lose an opportunity to earn some “extra” money. Very typical entrepreneurial way of thinking. This interest probably does not overweigh the other interest at stake, such as freedom of choice, freedom of expression and privacy, which are likely to be infringed if ISPs will control their network rigorously.

Whether the means of control, i.e. discrimination, is rationally connected to the objective? – if network management would eliminate network congestion, then, I suppose, it is rationally connected to its objective. However, if those ill-motivated users will nevertheless find other ways to congest the networks, regardless of ISPs’ network management, then I think it not rational to infringe others freedoms to achieve something that can be easily lost again.

Whether there is a minimal entrenchment on the rights? – taking into account that Deep Packet Inspection (DPI) analyzes all data going into and out of a computer, I think there can be nothing more entrenching to person’s privacy than DPI in our computer – dependent society. The entrenchment is not minimal.
Whether there are any alternative, less entrenching ways to achieve the objective? – According to the CTRC forum website, there are less entrenching methods to deal with Internet traffic available for the ISP. So-called “the best effort to all” approach presumes that all users are treated the same, and to accommodate growing internet traffic, the ISP will need to build enough networks. This certainly requires more spending by the ISP. According to case-law regarding balancing between a Charter right, such as right to privacy or freedom expression, suggest that the fact that respecting those rights may cost the infringer some extra money is not enough to justify the infringement of these rights.

This very brief analysis demonstrates that it will be very difficult for the ISPs to justify the infringement of our rights. If proposed Internet traffic management fully satisfies the above noted principles, I would not mind. However, ultimately we will have a result which will have to be corrected over the course of years to come through legislation and judiciary.

I believe what the ISPs are trying to do now resembles to a situation in which hunters burning the whole forest in order to kill the wolves. This is not a correct method. There might be some innocent parties who might be negatively affected during the process. Therefore, instead of using all-inclusive methods, it is rather more acceptable if they proposed to use narrower and more precise methods.

[updated 2009-04-05 16:18]

When I purchase an internet service, I expect it to be as fast as advertised, 24/7, regardless of the comunication protocol I use. I am appaled by the concept of "Service or application-level management". It will lead to abusive and anticompetitive actions from the ISP's. I find the tiered offerings to be acceptable, but they should offer more tiers to accomodate all users. The current 3 or 4 options that are generaly offered for residential use are inadequate. They have price points that are designed to force people to purchase services that are more expesive than what they need.

[updated 2009-04-05 17:00]

I favour subscriber-level management. My ISP (Shaw) sells several tiers of service, so users who require massive bandwidth can pay for it and the rest of us can pay less for the relatively small ammount we download.

I am strongly opposed to application-level management in any form. Most ISPs are in competition with several internet applications. Allowing them to control the bandwidth those applications receive will only restrict the services available to Canadian internet users.

Sell me my bandwidth and let me choose how to use it.

[updated 2009-04-06 21:53]

In my opinion, the real problem in this issue is the underlying physical network that is being managed. The existing networks are only limited in their physical bandwidth capabilities, thus, with every new generation of applications, more and more user bandwidth will be needed and ISPs will have ‘legitimate‘ reasons to manage this ever growing traffic. I am not favouring these Internet traffic management measures – I actually think the way these are implied are mostly impudent, as – imho – the utilized technologies are very sensible and powerful tools which were supposed to optimize user experience and welfare, but are more and more used to simply create revenue.

I think that fibre networks for instance would not necessarily require other management practices than Network over-provisioning or the Reliance on standard protocols, as the possible physical bandwidth capabilities are – at least at the moment – high enough to allow the maximum experience for the user without the necessity of throttling or shaping the bandwidth as it is done at the moment.

Until such a transition is made, of which I am unfortunately not convinced in the near future, I guess it will be hard to convince ISPs to use management tools in favour of the user as user in the first place, and not as revenue income source.

[updated 2009-04-07 09:49]

Internet traffic needs to be managed and as the Heavy Reading report points out, making the pipe bigger to accommodate more users is not a long term option. The issue we as users should have is that the intelligent traffic management technologies can be used by the ISPs to kill off their competition and redirect our traffic. For example, in a congested time priority should be given to VoIP over e-mail because it is more sensitive to latency and jittering. However, no ISP who also sells home phone will EVER do that. It just doesn’t make good business sense. Why encourage customers to go to the competition? Internet traffic needs to be managed by an unbiased party, not the ISPs.

[updated 2009-04-07 09:57]

The best effort for all approach- seems to be generating a wide consensus- as it doesn’t hinder or discriminate types of traffic or users- rather the best for all. Another aspect of network management that’s needs to be considered is forcing to increase competition among networks- pushing new ISPS into the Canadian market- ultimately to lower cost of services- which are insanely high. As well- why not operate the most available bandwidth possible- instead of creating a new generating of higher bandwidth in a few years or decade? The bandwidth is there- but these ISPs are screwing us up by minimizing it for us- to further profit.

[updated 2009-04-07 10:57]

Going forward, I see the managed approach to handling internet traffic winning out because the brute force method of providing as much bandwidth capacity won't be able to keep up with the sophistication of the applications being developed and deployed. That threshold will have to keep on increasing as the applications are released and so why not look into an intelligent method that is more robust and sustainable in the long run?

I condone the tiered approach to network management where customers pay for the level of bandwidth they wish to use, and are capped there. There would be strong penalties for exceeding it. This way, those users who simply browse and send email aren't subsidizing the more heavy users by having to pay the same amount to support their habit. If a users wishes to create loads of upstream traffic, let him/her pay more.

This does bring about the discrimination of network traffic which goes against he TCP/IP model, but at least the anonymity concept would hold. Clearly I don't condone DPI (nor does the FCC as they forced Comcast to stop during its short stint at this technique). There are alternatives to DPI, such as the tiered approach, and in my opinion, is the only scalable solution to handle future demands.

[updated 2009-04-07 11:12]

The best approach to internet traffic management is to ensure competitive ISPs that fight against each other for each consumer and to maintain each customer through both services and price.

If enough competition exists, consumers will dump ISPs that resort to throttling to address "congestion" or usage caps with high $$$ overages. If consumers are unable to switch to other ISPs because all ISPs mysteriously decided to go down the same billing path at the same time, this suggests market failure and non-competitive ISPs. It also suggests gross mismanagement of resources on the part of ISPs if the congestion is real (but I contend that in the absence of reliable evidence, this congestion dose not exist).

In other words, the best approach to internet traffic management is to encourage choices such that consumers can actually vote with their feet. If there are enough providers that are sufficiently competitive, at least one will have its cards in order and pick up all of the consumers unhappy with the incumbents, forcing the incumbents to provide a better service at a better price.

When the incumbents are moving toward providing less service at a higher price, this suggests market failure and a lack of competition in a market.

The main issue is that of last-mile providers and backbone aggregation links (which are strictly data services, not internet services) which are often owned by 1 provider, which also has retail internet services, but also same-ownership subsidiaries that could lose profit if the internet gets faster and more... usable.

The CRTC, in ensuring market-based internet access and therefore market-based internet traffic management that is in the benefits of both company and consumer, must ensure open access to these last mile services.

The CRTC should look at other countries that have regulatory systems that allow for competitors to access these last-mile links, both at the data layer, and at the physical layer. The CRTC should find out what works for consumers in other parts of the world and look to implement similar strategies.

The GAS and associated backbone aggregation tariffs have gone a long way in Canada toward enabling this type of market-based internet traffic management where the cable plant owner receives a significant fee for use of their equipment, the ISP actively markets and profits from the service they provide after paying the cable plant owner, and the consumer benefits by getting the service that they want. But the cracks in GAS have been appearing, with Bell degrading GAS-based providers to its own levels and Bell going as far as parliament to overturn an Open Access decision by the CRTC. GAS needs to have more teeth, and more limits on what can be done on a given paid-for link.

GAS provides for the type of market-based internet traffic management that would serve Canadians best. GAS and associated services should be mandatory options available for coaxial-based providers as well, but also for other built-out services such as fibre and DSL remotes, for they suffer from the exact same issues as copper-based services.

I suggest the following as reading material regarding this type of market-based approach to internet traffic management:
http://ecommmedia.com/blog/2009/04/brough-turner-structural-bypass.html
http://link.wits.ac.za/papers/LINK.pdf
http://www.dslreports.com/forum/r22195751-Re-In-Japan-its-160-Mbps-service-costs-6000-yen-60-per-mon

[updated 2009-04-07 18:59]

I think that the best approach to use to manage Internet traffic is "The Best Effort to All Approach", since it is open, fair, and transparent.

1) This method relies on well documented protocols for managing network congestion, so their is transparency and predictability in network behavior for application developers.

2) Network capacity limitations are also immediately felt by users, thereby allowing customers to clearly gauge when ISPs have over-sold the network. The managed approach masks network capacitiy problems from customers so it is difficult for customers to judge between good and bad ISPs, with regards to network capacity.

3) It creates clear competitive incentive for ISPs to continue to invest in network capacity.

4) It avoids potential privacy concerns that the Managed Approach creates.

I also do not feel that demand for increased network capacity will increase endlessly, but instead will plateau at a certain point. There is only so much data, just like any other good, that any one household can consume. Once this point is reached, demand for increased bandwidth will subside.

[updated 2009-04-07 21:02]

it makes more sense to me that the isp's just allow the traffic to flow as it does without interference.some will use more than others if the isp's want to build there customer base the should do it the old fashion way earn it by giving the best service,increasing and upgrading there equipment . this would give them the edge instead of ripping off the customers they have by charging more for what there getting already without upgrading the system.

[updated 2009-04-08 01:26]

In reply to the Cisco findings, I have to disagree somewhat. With the advent of p2p, many users instead of downloading video multiple times, download only once. Sometimes they may share with their friends afterwards. Comcast for the longest time flamed bittorrent, claiming it was a bane on their existance... then suddenly, they had a change of heart. Why? Youtube. Youtube made it possible for people, wherever they were, to share a video easily, effectively, and the only people that really paid the bill for that was the people who provided the service.

Granted, as the cisco finding points out, bandwidth DOES cost money. The issue is not THAT it costs money, but HOW MUCH does it actually cost? Herein lies the problem. The actual cost vs the cost that one is stating they're willing to sell it for. If you assume that a dollar per gigabyte is not excessive, and people, for example, decide to download a high definition show that is a gigabyte in size, then people paid 1 dollar for that show. That may not sound like too much to ask, but this would also put itunes and amazon unbox, and many other services farther into the ground. The real cost of a show, so-called legally, will then be 1 dollar for the bandwidth, and probably 4 dollars for the show to download it from itunes. The problem is, all of this is an inflated market. A gigabyte doesn't cost $1, it costs on average, 1-6 cents. Even a 100% markup on the price of bandwidth gives you a range of 2-12 cents per gigabyte. Consumers know the cost of bandwidth, bandwidth is not the issue, the issue is the unreasonableness of the cost. If Cogent can give data to the average joe for around 400 cents per 320 gigabytes (but in limited locations) why are the average consumers being charged so much? Bell's own network infrastructure overhead costs can't be 8000% or more, can it? That just sounds like an insane amount of money to connect a network, and if that's really the cost Bell has to pay for internal bandwidth, then it deserves to go under.

[updated 2009-04-08 08:07]

There is a lot of dark fiber under Toronto right now that is not lit up and I suspect it's the same in other cities and possibly even towns. If they used what was in place already (properly) I suspect there would be enough bandwidth for everybody to steam multiple high bandwidth sites at the same time.
I'm in favour of over-provisioning though.

[updated 2009-04-08 22:02]

The best approach towards internet traffic management is to provide for greater scrutiny regarding enforcing the Telecommunications Act and Broadcasting Act against large corporations such as Bell Canada. Such corporations currently hold a significant monopoly in terms of providing telecommunication services to the general public. Therefore, when claims demonstrating that Bell would gain a significant competitive advantage from engaging in anti-net neutral techniques, such as throttling, the CRTC should analyze the situation from an objective standpoint and determine who would gain a financial advantage from these methods.

Net-neutrality with the internet is threatened by corporations such as Bell, who decide, without appropriate oversight, what applications will be “throttled” or rate-limited.

The negative aspect of a “managed approach” towards internet traffic management is that it is prone to discrimination. Specifically, corporations such as Bell and Rogers, stand to acquire a financial advantage by “throttling” or rate-limiting users who use competitive technologies such as Skype to communicate over the internet instead of using Bell or Rogers telephone services. If such users consistently receive slow connections when using these technologies then they are less likely to continue to use them.

Another negative aspect of a “managed approach” by corporations involves privacy concerns. Specifically, a corporation’s ability to use deep packet inspection in order scrutinize individual packets of data and de-prioritize the content, is clearly contrary to section 27(2) of the Telecommunications Act. For example, an individual such as a filmmaker would clearly be disadvantaged by throttling if Bell used anti-net neutral techniques to determine, based on their criteria, that a filmmaker’s content is not a priority.

In the recent CRTC decision 2008-108, Bell Canada claimed that its DPI technology “examines the packet headers, not the contents, of communication exchanges”. Whereas DPI technology has the potential to examine the contents of information packets, only shallow packet inspection (SPI) technology extracts basic protocol information such as IP addresses (source, destination) and other low-level connection states. Therefore, the CRTC should mandate that corporations such as Bell utilize SPI technology instead of DPI technology because of the potential for abuse contrary to section 27(2) of the Telecommunications Act and privacy violations contrary to the Personal Information Privacy and Electronic Documents Act (PIPEDA).

In terms of section 36 of the Telecommunications Act, the CRTC stated that “[it] considers that the traffic shaping carried out by Bell Canada does not influence the meaning or purpose of telecommunications.” However, if someone is altering the timing of the content and degrading its overall quality, the meaning of the content will be lost. This is especially true with respect to peer-to-peer technologies such as Skype and other streaming media, in which delays affect the meaning of the communication. Although Bell claimed it is not “exercising any editorial control of the content being transferred”, targeting specific applications for throttling is exercising editorial control over the medium containing the content. Therefore, I believe that by targeting peer-to-peer applications with throttling techniques, Bell is indirectly exercising an editorial control over the content of communications per section 36 of the Telecommunications Act, by discriminating against the medium of communication (ie. peer-to-peer applications).

[updated 2009-04-09 11:52]

Application-level traffic management is NOT a prudent approach.

The ISPs are not entitled to decide which Internet applications to prefer over others. The network and the customers are theirs, but the applications are created by the IT industry.

Will the developers have to get approval from the ISPs to ensure unthrottled function of their applications on the Internet?

ISPs' discriminatory influence over Internet applications can undermine innovation and consumer-choice in that sector.

[updated 2009-04-11 00:16]

Disclosure: I work for a telecom equipment provider company, although I don't work with the equipment handling DPI.

This has become cat and mouse game: Telco companies trying to filter p2p traffic will only succeed for sometime, while the p2p overcomes -again- the new filtering. For what we can foresee, this game will *not* end soon. So ... Its a waste of time and effort for both sides to try to filter and overcome these limitation.

Contrary to many people believes, overprovisioning a network its not the more efficient way with dealing with this problem for one simple reason: It costs a lot of money! Specially since the Internet is constantly creating applications that consumes a lot of bandwidth far away for the initial networks plans.

To put it in other words, it cost a lot of money to keep up with the exponential growth of traffic demand of the internet. Its easy to say that the ISP pay for it, but in the end we (the subscribers) will pay it.

The problem is that the business model the ISP are using *became* the wrong one.

Previous experiences

The electrical utilities solve this problem years ago:

If I am a electrical residential costumer I will be charged accordingly, but if suddenly I start to use a heavy electrical motor @ home, which spikes the load I'm putting on the electrical network and if I continue this operation for certain amount of time (lets say two months), for sure the electrical company will come to me and say: We have to start billing you as an industrial customer, since you're using the electrical network as such.

This is the way the electrical industry has worked since we know it. We are free to use the electricity as we please, but will pay for it! It can be argued the utility charges you more because it has to generate more of it, true. But generally, this its not the big problem, the problem for the electrical companies is transmission (*the network*), and as we know, if we overcharge the electrical network then we experience blackouts. In the case of ISP, it doesn't cost them a dime to produce the info, but it does cost them a lot to transport it at high speeds, specially core links.

In case you're wondering if laying of cables is a significant cost: It is! Think of it of all the dealing a company has to do with communities expanding the network (laying out cables), and you'll know why there aren't more ISP with their own infrastructure around.

So, what to do?

Giving the current situation, I think Subscriber-level management is the way to go.

We can start with a very simple measure: The average of GBs consumed per subscribers. Lets say -for example- a subscribers use 20 GBs monthly.

Then we can create three tiers:

- Those lower than 20 GBs will pay less, let say 15 CAD.
- Those on average pay 25 CAD
- And those over will pay per GB, either from the beginning or after the average is consumed.

Since we don't know what new applications the future have for us, CRTC will revise the average consumed per household, and if necessary raise it, so we can continue to enjoy the progress without degrading our pockets :-)

[updated 2009-04-11 12:01]

As a internet consumer and a person who works developing telecom equipment I view the internet connection to my home as a pipe. I a willing to pay for defined service level characteristics but I dont accept that an ISP should - In Any Way - be allowed to dictate what applications I use over that pipe.

I view any system based upon, deep packet technologies, as a profound violation of my privacy, and a technically flawed and unnecessary approach to traffic management.

I expect the CRTC to strongly protect the public interest by insuring net neutrality.

[updated 2009-04-12 13:12]

As soon as I hear terms like "management" I get a little nervous, and a little resentful that I am going to be "managed". Either of the two scenarios beg the question: "Who gets to be the Hammer of Thor?" In other words, if I use P2P protocol to get an episode of "The Office" that was pre-empted by football, and I pay for satellite TV, and I pay for my Internet connection, and I have to accept broadcasts of football even though I'm not a fan....how is it even relevant that An Authority can decide that, no, after all, they don't Approve of p2p, gnu, BitTorrent...etc....?

In much the same way that I would not accept the phone companies monitoring my calls and deciding that I shouldn't be allowed to call Iceland or Luxembourg or my neighbour, neither do I think any self-appointed "governing" body should be able to arbitrarily manage and monitor my digital communications.

[updated 2009-04-12 13:37]

Any ‘Managed Approach’ system that deviates from the public’s unobstructed access to information on the internet is a discriminatory and an unfair business practice. One of the CRTC’s roles, in regards to the Telecommunications Act, is to ensure that in section 27(2): No Canadian carrier shall ... unjustly discriminate or give an undue or unreasonable preference toward any person, including itself, or subject any person to an undue or unreasonable disadvantage. If the CRTC allows ISPs to withhold and to give preference to selected data packets, they will be allowing ISPs to discriminate against the free enterprise and the will of the public.

It is the CRTC’s role to preserve Canadian culture, not by restricting our access to knowledge abroad, but by providing freedom and accessibility to our culture. For our protection, the Telecommunications Act explicitly outlines its objectives in section 7, and there are several identifiable points in this Act that illustrates a Managed Approach system would simply be unjust. Although, the CRTC must distinguish the future of ISPs without breaching our public law and by conforming to some of the following ‘objectives’ of the Telecommunications Act, section 7 instructs ISPs to:
(a) ... enrich and strengthen the social and economic fabric of Canada and its regions;
(b) render reliable and affordable telecommunications services of high quality accessible to Canadians in both urban and rural areas in all regions of Canada; ...
(g) ... encourage innovation in the provision of telecommunications services ...
(h) respond to the economic and social requirements of users of telecommunications services.
The CRTC must ensure that any form of regulation does not breach the Telecommunications Act, or economic and social harm will ensue. The CRTC, too, must be reminded that competition amongst ISPs will encourage innovation, and that competition will improve accessibility to the internet for all Canadians.

It is important to recognize the deregulation of Bell’s telephone monopoly in 1997 improved telephone services throughout Canada and helped make telephone communication more accessible; even though Bell warned Canadians that the service quality would diminish after the deregulation of its monopoly—it subsequently improved due to competition. Deregulation was the mandate that forced Bell into the fair business practice of allowing other companies to compete for customers by sharing the same infrastructure. To follow suit, Japan also forced broadband companies to allow their competitors to share the infrastructure. As Susan P. Crawford writes in her article “Network Rules”, Japan’s internet service improved; became more accessible, and less expensive. She also notes that in Germany, where a monopoly is held by Deutsche Telekom, the internet service is unaffordable and slower in comparison to other developed countries. Competition is healthy for the public and the business sector. Competition will help ISPs become more innovative to the benefit of the ISPs and their customers. Although competition is important, ISPs should not have any opportunity to discriminate against their competition.

It is quite clear that if the CRTC allows ISPs to censor and selectively withhold packet information, the internet could become a failed vital organ in Canada. If ISPs are able to discriminate against any competition and to drive up internet service costs, the overall quality of internet service will be diminished. A diminished internet service will hurt Canadians in many ways, but it is important to recognize the damage will: 1) limit e-commerce; 2) obstruct access to information; 3) create a corporate driven censorship that unfairly discriminates against competitors. The internet has become as essential as highways in Canada, and without unobstructed access, our freedom to information would not be limited in the same manner as those who live in communist countries, such as China or Cuba, with restricted censored access. If the CRTC does support a Managed Approach system, they will be preventing any innovation of our infrastructure by allowing ISPs a diversion from their current responsibility to provide the service their customers have already been paying.

[updated 2009-04-12 16:06]

I, for one am appalled that the CRTC would consider allowing the sole provider of connectivity to the entire DSL user community in Quebec and Ontario to engage in traffic shaping and the imposition of traffic caps for subscribers.

More interesting is that by the wording of Bell Canada's application, it appears that Bell would exempt their own end users from such caps, while imposing them on the subscribers of other ISPs, giving Bell an unfair advantage over the competition.

Further – As a DSL subscriber, what I lease is a circuit, not traffic. I should have the right to send whatever I like in both directions and the network provider has the responsibility of providing adequate capacity to carry said traffic;

Besides; Bell appears to be operating under predatory practices; limiting the connectivity of subscribers to competitive ISPs – in my own case I was informed by Bell technicians that my DSLAM was not the ideal one for my location, but that Bell would not permit them to relocate my connection to the appropriate one because I subscribe to an 'alternative ISP'.

Other subscribers have seen their connection speeds drop from from 5-7 megabits to less than 512 Kbps in some cases – these are not isolated incidents, but rather appear to be a almost standard policy practiced by Bell – in my own case I saw my connection from from 5 to 3 – I did manage to get Bell to increase it to 4 megabits – not what I’m supposed to get, however it is tolerable.

At 5 megabits per second, one could theoretically fully use a 60 Gigabyte cap in about 30 hours if one transmitted continuously . . . and, I’m told that Bell would like to be able to disconnect a subscriber who exceeds 300 Gigabytes of transfer (admittedly a lot). But why the caps at all?

Bell already benefits from their ability to concentrate user traffic – now they throttle bandwidth because in theory their front end would overload; yet it all self regulates in the end. Bell needn’t throttle traffic. The traffic self regulates simply bause when there is a bottleneck, only ‘so much’ traffic can get through – and the distribution of that traffic is random – so everyone should be equally affected by this ‘natural throttling’.

However Bell SHOULD be building adequate capacity to serve their entire client base . . .

Bell does collect almost $18 from the ISPs they service and as much a3 times as much from their subscribers; I’d be willing to wager that Bell already makes a lot of money from operating their DSL front end. To allow them more would be promoting profiteering. To allow them to establish traffic caps and to throttle traffic would be almost sinful;

Bell should *not* be able to treat their subscribers any differently than they do those of the alternative providers. Special DSLAMs, no caps and other benefits should be disallowed. A level playing field is required, competitively-speaking.

Why is it that Americans don’t have Caps and their ISPs don’t throttle, yet their population density on the net is much greater than ours in Canada.

Please do not pander to Bell’s rediculous assertions.

Respectfully

[updated 2009-04-14 17:31]

First of all, I think the companies (specifically Bell and Rogers) has not invested a lot in internet infrastructure. Canada has drop majorly in internet speed comparison to other developed countries. We have been in the same range of speeds for years and the costs of internet has not dropped. Bell decided to put money into machines that manage traffic. When that same money could of been made to improve the network. Traffic management is now used to limit high bandwidth applications to squeeze more people into the same node and again give them another excuse not to invest. And now there are plans to limit bandwidth with caps and again another penny-squeeze to cram more people into outdated machines.

I believe that these companies must seriously invest into their infrastructure and provide competing speeds or Canada and all internet based Canadian business will be left in the dust. Why has their been no Canadian-made internet video streaming service? No online gaming company? No other internet Canadian-made service? The internet is growing and will continue to grow. New technologies are currently being made and costs are going down. Video streaming is the current thread now but what is next new thread?

The cost and effort used to manage the traffic is a waste. Like for example, we limit P2P. People move to something else like video streaming. We then limit video streaming. So people move to direct downloads. And then we limit that. And then the next and the next. Eventually, traffic management will limit all methods and protocols to the speeds of a 56k modem (remember dial up?). And what we do then? With nothing else to limit, what other "management" can we do? Nothing. Traffic management is only a temporary delay that will reach a dead end when used to the fullest extent. Traffic management is a step in the wrong direction. It is going backwards, not forward. With other countries investing in their infrastructure, Canada will be dead last place if continue this path and we won't be able to compete in the internet world.

So a best effort for all approach is the best way and only way to go.

[updated 2009-04-14 23:34]

In a Managed Approach, differentiation soon leads to discrimination. As other people have commented, this system is easily abused and will be abused as long as we have a telco monopoly structure in Canada.

A "Best Effort" Approach has the benefit of not restricting service for any individual and if telcos had to conform to this standard it would provide an incentive to modernize Canada's infrastructure (I hope).

[updated 2009-04-15 11:52]

The Best Effort To All approach is the best approach to internet traffic management because it upholds the idea of net neutrality the best. Network over-provisioning may be expensive for ISPs, however that is what they are paid large sums of cash each month by each of their subscribers for.
The Managed Approach, as we have already seen, can be abused by the ISPs to stiffle competition and innovation. If ISPs feel they absolutely must manage internet traffic, it should either be network-wide or subscriber-level. Service or application-level management infringes on the rights of the subscriber to use the protocols they want. As it has been said before by many people, what protocols I use on the internet are none of the ISPs business. That is a privacy issue. They should not have the right to decide which applications they will allow. They are paid big money each month by each subscriber to provide a connection to the internet - that is all. They do not have the right to tell me which protocols they think I should be using by limiting my ability to use one program over another.

[updated 2009-04-15 12:53]

An ISP should never manage or restrict the internet. If a company’s chooses to be an ISP then they should conduct business in a way that conforms to the idea the internet is not theirs. The internet is a public pipe and is owned by nobody. If the internet in Canada is "managed" Canadians freedoms will be eroded.

An ISP should never restrict the internet. If a company’s chooses to be an ISP then they should conduct business in a way that conforms to the idea the internet is not theirs. The internet is a public pipe and is owned by nobody. If the internet is "managed" Canadians freedoms will be eroded.

The CRTC should ensure Canadians are provided with a way to access the internet as the internet has become an importent way of telecomunication for the whole world.

[updated 2009-04-16 03:27]

Bell Canada introduced throttling to Sympatico customers back in '07. I disagree with their decision to do so, so I left and went with another provider. At the time Rogers was my cable company so I could have chosen to go with them, but they were already doing this. So, I found a company called Teksavvy that did not throttle content, and I chose this company to do business with.

I disagree with Bell throttling Sympatico customers but feel that it is their right to affect THEIR customers in anyway they want. If they want to provide a certain type of service fine. I disagree with Rogers throttling their customers, but again, they can run their service the way they want. Unfortunately, when you use a cable company as your internet provider you are trapped within their monopoly. No one else can offer you cable service except your cable company.

When you are a DSL customer you have choices to who you deal with.

With Bell now throttling ALL internet connections on DSL they have effectively removed all choice. They have created a disincentive to move to another company because you will in effect receive the same level of service. This move can ONLY be seen as anti-competitive.

I purchase a connection to the internet. A connection that I can use for any legal purpose I see fit. I deal with Teksavvy because that is the company I choose to deal with. I am forced to use the 'last mile' from Bell Canada because we as Canadians have given Bell the ability to run their cabling everywhere. Are we going to give EVERY company the ability to do the same? Is my street going to have 16 more rows of telephone poles added so that every provider that operates in Canada can run their own 'last mile' Of course not.

So why do we accept the argument from Bell that they should be able to manipulate the service of whomever they want?

What is next, slowing or blocking video traffic because Bell wants people to use its satellite service for television? Slowing or blocking Skype and other VoIP providers so that we have to use its phone lines. After all their argument is 'The network belongs to them, and they can therefore behave how they see fit'.

This argument is totally without merit and this company should be fined for interfering with the normal business practices of their competitors.

I urge the commission to reject Bell's arguments in favour of traffic shaping of its competitors.

Howard Rabb

[updated 2009-04-16 06:43]

ISP throttling has been a bane to consumers for years. Upstream bit rates are lock at ridiculously low rates. For example my ISP (Shaw) recently pitched 50% faster, only for downstream, upstream was not changed. That is misleading from a legal standpoint. I would like to see upstream=downstream like it was in the dial-up years.

The abuses inhibit web cams as well as P2P which is legal in Canada.

My current plan has 25 Mb download but 1 Mb upload. Sure makes P2P slow and web cams choppy. Makes it hard to watch a girlfriend do something like demonstrate an idea to make a salad.

[updated 2009-04-19 15:11]

In Yukon Territory, the ISP has a monopoly. They also provide telephone land-lines. Allowing a company with a monopoly on providing internet services to protect their interest in long-distance by throttling anyone who tries to use voice-over-internet or video-over-internet -- users who have paid for service at a certain rate -- is bad practice.

Allowing such management techniques as selective management of any sort opens the doors for just such misuses of the technology. In an area where the infrastructure could allow every client of the local monopoly ISP to run quite comfortably at the highest available speeds with little to no change in service for anyone, throttling back paid-for service selectively is malicious. Companies in the Yukon have shown, in the past, that they cannot be trusted to NOT use selective throttling of bandwidth maliciously in order to maximize their short-term profits rather than manage a vital resource for everyone.

"Net neutrality" is needed for the industry, and for fairness to people who are paying for a service in good faith. Any use of selective technology is open for great abuse, and should only ever be used, if at all, in cases where an immediate concern calls for their use. However, allowing such use would only create an incentive for short-sighted ISPs to supply LESS bandwidth so that they could justify use of selective management tools to maximize profit through abusive practices.

Thank you.

[updated 2009-04-19 18:27]

No Traffic Management at all is the only way to go. As a consumer I have paid my ISP to expect certain speeds. If I am not getting those speeds, they are in violation.
It cannot be allowed to continue.
I do not pay over 50.00 dollars a month so that I can get the same speeds my 56K modem used to give me.
CRTC, you ought to be banning all traffic throttling, and making Bell and Rogers live up to their advertisements of speeds.

[updated 2009-04-20 07:16]

I don't have a problem with reasonable network management practices. Here's my problem: all of the management practices listed above allow ISPs to collect more money from subscribers while providing less service. ISPs have been overselling their capacity for years, and with network management approaches that allow additional fees to be charged, ISPs have a strong incentive to continue selling the same limited resource to more and more subscribers.

Overselling is a legitimate business practice, but when it's done with the knowledge that there's no way a subscriber could ever be provided with the service advertised, it's fraud. Some degree of network management is probably reasonable in the short term, but there has to be oversight and control on just how much overselling is allowed, and a recognition that ISPs MUST expand capacity to meet their service commitments or clearly specify limits to customers before they enter a subscription agreement. Any and all traffic management practices in use must be explicitly specified in advance, and they must be temporary measures, not a permanent business practice.

[updated 2009-04-20 08:05]

More competition. For many subscribers, conventional telcos control Internet access through their land-line networks. The solution to traffic management is a more competitive ISP market, utilizing a broader array of technologies, including wireless and satellite. Let market forces and innovation set the standards for bandwidth and throughput.

[updated 2009-04-20 08:11]

The best approach to traffic management is to not have any traffic management. Invest in newer and better technology. If Canada doesn't invest in internet technology, it will fall further behind, and it will lose a lot of business to many other countries that are building a strong internet technology infrastructure and seeing the benefits of it. If Canada wants a future with a strong technology market, it needs to invest in the hardware, and stop companies like Bell and Rogers who just want to hold things back.

[updated 2009-04-20 08:34]

Since ISP's are primarily concerned with the bottom line and not service provision they should be forced to adopt the "best effort to all" approach. If this requires investment in network improvements the ISP's will undoubtedly find ways to recover the costs.

[updated 2009-04-20 08:47]

I think the best approach is whatever a given company can do without irritating its customers. If a company can find a mix of services, management policies, and price that satisfies a market, then good for them. Keep it up.

[updated 2009-04-20 08:56]

If ISPs cannot afford to provide unlimited bandwidth at the advertised speed without traffic shaping, they should not be selling it. I would be willing to go back to a capped bandwidth arrangement to avoid sacrificing net neutrality (although I sincerely hope that the stronger competitors will be able to continue offering unlimited bandwidth so I have one less quota to manage in my life). If capped bandwidth is the only way to provide content-neutral Internet service, ISPs should at least have the courtesy to offer (1) online tools to check your current usage, (2) email notifications when you are approaching your quota, and (3) the option to buy more bandwidth at the same cost.

[updated 2009-04-20 09:18]

Let's imagine traffic management techniques in other industries to put the internet in perspective. Imagine if trucking companies owned the highways and restricted traffic except for those vehicles that paid a fee for a fast lane while everyone else was deliberately slowed down. The problem would not be that drivers would be annoyed, the problem would that it would restrict trade and commerce. It would be bad for all, bad for the economy. That is my biggest concern, not my individual speed, but our ability to conduct commerce, innovate and use the internet to grow the economy. If we have a restricted system and the USA, Europe and India does not, we will fall behind technologically and it will be more difficult to compete. We cannot, CANNOT have a backward, throttled, censored (that is what it is) system here while the new administration in the USA goes forward with a fast, free and open system that encourages an entrepreneurial, innovative atmosphere for economic development. Where will the 21st century jobs go? I care more for economic growth for all Canadians than I care for Bell's profit margins. That should be the main concern of the CRTC.

[updated 2009-04-20 10:29]

Absolutely the "best effort for all" approach. An ISP should not have any editorial power into what data, and what amounts there of, can reach the end-user.

Moreover where there is a clear financial motive for preferential treatment to their premium services, to allow anything else, would be anti-competitive and could stifle innovation.

If we begin allowing the ISPs this control over content for some of the current "boogeymen" such as bittorrent, what is to stop them from applying those rules to new protocols or web-services that have not yet been developed. This is especially troublesome as the revenue from traditional broadcast media declines.

The bottom line is that an ISP should not be empowered to do anything more than provide unfettered internet traffic. If they wish to sell premium content and services through the World Wide Web they may do so like anyone else: through innovation and competitive business practices.

[updated 2009-04-20 10:48]

Lets look at these wonderful choices, shall we?

The Managed Approach

- Network-wide management: Giving all consumers less with no benefit to anyone.
- Service or Application-Level Management: Internet Racism, only instead of discriminating against race, you discriminate against types of users.
- Subscriber-Level Management: Cash grab

The Best Effort to All Approach

- Network over-provisioning: Actually providing customers the capacity to get the service which they pay for. What a radical and new concept!!
- Reliance on standard protocols: This actually should be a part of the above. Most internet protocols do have some form of fail safe built in. For example, in the case of bittorrent, a user can only download as fast as other people can upload, and can only download for as long as the file exists (You can only download 700 mb of a 700 mb movie). Hence, although I may achieve a fairly high speed, it likely will not match HTTP downloading.

[updated 2009-04-20 10:50]

Going solely by the title "best effort to all" leads one to choose this over "managed" approach. But titles aside, the best effort to all is the way to go in my opinion. As we continue to modernize, the internet is more and more becoming the hub of contemporary expression. It is a necessity and in order for it to reach the masses, on an equal basis the managed approach does not suffice.

[updated 2009-04-20 11:15]

Speed limits have nothing to do with congestion, volume/bandwidth is the deciding factor. Just like real road traffic, if there is too much volume then the entire system slows but at least it is equally spread out among users. Compare highways with speed limits and those without, the congestion factor is similar.

Bandwidth throttling is therefore not an effective way to deal with internet congestion because it discriminates one person over another and the only real way to deal with volume is changing bandwidth capacity.

[updated 2009-04-20 11:18]

It's ridiculous to bend to the whims of our internet providers. The trend should not be limiting our access as technology becomes more sophisticated, but to have our providers keep up with the times. They are businesses that are lazy and lacking innovation. A best effort to all approach is the only one that makes sense when dealing with the provision of our internet services.

[updated 2009-04-20 11:36]

Thesis excerpt- Bandwidth is Political- Nancy Paterson - YorkU

Canadian Radio-television and Telecommunications Commission Chairperson Konrad von Finckenstein introduced Net Neutrality in his speech to the 2007 Broadcasting Invitational Summit and again at the 2008 Canadian Telecom Summit. He stated that the Canadian Radio-television and Telecommunications Commission's New Media Initiative will include "striking a social, cultural and economic balance to deal with Internet traffic prioritization" (von Finckenstein, 2008). Traffic prioritization or traffic shaping is only a part of the entire concept of data reachability within the concept of a neutral network. A neutral network must ensure as a best practice that data is reachable. What happens if you traffic shape the internet data packets down to zero? There is no traffic exchanged.

Shaping packets of Internet traffic is termed ‘Quality of service’ by the commercial networks who practice it. But shaping packets is part of a continuum of tools for policy based control of Internet traffic flow. The continuum of Internet traffic control is as follows: packet filtering which includes disallowing packets as well as traffic shaping data packets for a specific single (or multiple) Internet address; route filtering through prefix lists which aggregate a single block, or blocks of Internet addresses to be disallowed (McPherson, Sangli & White, 2005, p. 84, 185); autonomous system path filtering, which is the practice of disallowing routes with a specific autonomous system number in the path (Keshav, 1997, p. 342). Lastly, physically cutting the wired links and de-peering with the disallowed autonomous system is the analog unplugging or severing of the connectivity. Route filtering generally takes place before autonomous system filtering in the order of operations in Border gateway protocol (McPherson, Sangli & White, 2005, p. 235).

The practice of filtering an Internet address for a server which hosts many different unrelated websites is seen in the well known example of non-neutrality in Canada which occurred during the 2005 Telus labour dispute. Telus, the familiar Canadian telecommunications company, blocked access to the Telecommunications Workers Union's website also blocking “more than six hundred additional websites hosted at the same IP address and cut off entire communities from the controversial content” (Geist, 2005, December 19). Telus blocked access to the pro-union website by blocking the server on which it was hosted through route filtering. Researchers at Harvard, Cambridge and the University of Toronto, OpenNet Initiative found that Telus’s actions resulted in exactly seven hundred and sixty six unrelated sites also being blocked (OpenNet Initiative Bulletin 010, 2005).

De-peering involves breaking the actual physical path between networks and unplugging the actual physical (wired) connections. Autonomous system number filtering is a separate operation from de-peering and usually accompanies the technical severing of the connection. The York University example below, details the effect of this scenario in March 2008 resulting from de-peering. A professor using the on-campus network (commercial Internet connectivity supplied by Cogent) attempted to reach an internet website located somewhere in Europe that was important to his research. After repeated attempts he could not reach the site which he had been accustomed to reaching in the past, so he contacted the Information Technology department at the university. They were mystified why this unreachability would be the case. When the professor went home however, he found that he could reach the website consistently. At his residence, the Professor purchased bandwidth from Rogers (and its upstream providers) which did not sever their peering relationship with TeliaSonera. After several days the Information Technology department found out that the university’s transit supplier Cogent had severed a peering relationship with a network in Europe called TeliaSonera which was the bandwidth provider for the website that the Professor was trying to reach. Cogent did not proactively inform the University of the issue and the loss of connectivity. The University had to open a trouble ‘ticket’ in order to receive an explanation from Cogent regarding the reason for the failure (Paterson, 2008, Apr 20). The de-peerng resulted in an outage that took down a significant part of the Internet according to the Renesys Blog. This blog is hosted by Renesys corporation which a commercial internet monitoring and reporting company providing high quality unbiased information to bandwidth stakeholders. The blog is a very useful source of information regarding peering and internet infrastructure. Earl Zmijewski wrote in his Renesys post titled “You can’t get there from here” that:

[T]he list of impacted networks is too long to be included here, but they include a wide range of commercial, educational and government clients. On the Telia side, the victims include the Swedish Defense Data Agency, the Finnish State Computer Center, and broadband customers in St. Petersburg. With regard to Cogent, Blue Cross and Blue Shield of Delaware, Kansas State University and Reuters America were all collateral damage (Zmijewski, 2008a).

In the 2008 Cogent-TeliaSonera de-peering, Cogent never informed York University that it was severing their relationship with a good part of Europe’s Internet by de-peering with TeliaSonera. De-peering in itself is not typically a sufficient condition to cause data unreachability but when the de-peering is accompanied by another step, autonomous system number filtering where one network began disallowing any data traffic from the other network through intermediate networks, then websites become unreachable due to arbitrariness in peering. To one high level network the other network simply does not exist.

Net neutrality and deep packet inspection are difficult to define. Once used to try to limit the impact of peer-to-peer networks through filtering or disallowing packets in Internet protocol networks, newer deep packet inspection platforms (read hardware) are assisting networks to control how they handle specific applications and scrutinize data traffic.

Multi protocol label switching (MPLS) networks implement deep packet inspection. Multi protocol label switching kind of replaces Internet protocol routing - packet forwarding decision (outgoing interface and next hop router) is no longer based on fields in Internet protocol header (usually destination address) and routing table, but on labels that are attached to packets. Labels are applied to packets which are then read/routed by multi protocol label switching network compliant routers. Multi protocol label switching network networks are, at present, inside IP networks but are taking them over. For example, AT&T's 'U-verse' is a multi protocol label switching network.

In Canada, traffic shaping practices were the subject of a complaint filed to the Canadian Radio-television and Telecommunications Commission in April 2008 by Canada's largest Internet service provider association, the Canadian Association of Internet Providers, against Bell Canada. The complaint by the association of local access Internet service providers requests that the Canadian Radio-television and Telecommunications Commission direct Bell Canada to cease throttling or traffic shaping its wholesale bandwidth interconnections with them. Michael Geist, a law professor at the University of Ottawa, details the specific argument:

Bell Canada’s traffic shaping measures discriminate between different types of content and the users of that content (contrary to subsection 27(2) of the [Telecommunications] Act). It is also evident, however, that these measures violate Bell’s obligation under section 36 of the Act not to control the content or influence the meaning or purpose of telecommunications carried by it for the public (Geist, 2008a).

[updated 2009-04-20 11:52]

I think that Best effort approach is the only right way to go. Internet became utility service over time and if everyone knows that rules are the same for everyone then everyone will do their job based on this assumption (eventually everyone will take it as law of nature if you will). Application developers will design their applications with assumption that there is no guaranteed speed, latency or reliability if they decide to use Internet as delivery medium and service providers will focus on extending their infrastructure to meet demands of their customers. Of course this will cost money, but this is business - if everyone plays by same rules and only the best survive then this is how we progress. The current practice when service providers can squeeze more customers on their networks without investing in infrastructure and partially degrading quality of services that THEY think are not important is completely wrong.

The managed approach can be easily justified by its proponents but let's look at bigger picture. We have only one Internet on this planet (it's like air or ocean), so imagine what power one will have if he or she can control it (imagine someone selling you air here). The managed approach gives companies with enough money exactly this opportunity.

This has started already and in the long run we could end up with dozen media conglomerates controlling the Internet entirely, suppressing any independent service providers and killing all competition. This will also cause service providers to focus more on serving their sponsors rather then their customers. One can speculate even further that with this much power few corporations will be able to influence government policies even start wars (remember Iraq), but fortunately for us Canadians there is CRTC that supposed to protect us from this course of events.

Also note that it is not Bell or Rogers who owns Internet - they simply provide access so they should play by rules common to everyone. The problem is that these rules are not cast in stone yet and CRTC could set a good precedent as well as show good example to the rest of the world.

[updated 2009-04-20 12:08]

All the approaches except "Service or Application-level management" seem to be fair. I am concerned that "Service or Application-level management" would lend itself to abuse by companies. If this is allowed regulation should specify a very narrow range of communications that can be discriminated against (I.e. if some virus is causing a huge amount of traffic, this traffic could be discriminated against).

I also have concerns that, without adequate competition, insufficient network capacity might be built in the long run. I'm not sure what the best solution to this is.

[updated 2009-04-20 13:01]

First...ISPs are privileged to be able to manage the internet service provision...the internet has become a backbone of our economy and as such should not be in the hands of private enterprise in the first place unless they guarantee fair access to all with over-provisioning. They should never be given the opportunity to control the flow of data as this leads to censorship and preferred customer practices which limit business freedom. They should only be given a license if they can guarantee maximum service availability and speed to all customers and all applications. If I were a small e-comm. business dependent on reasonable speeds for financial transactions over my web site and my ISP slowed my connection in favour of a higher paying or larger customer, it could harm my business.
We need serious regulation to protect all consumers equally.

[updated 2009-04-20 15:03]

The Best Effort approach is better because it pushes forward high speed internet access for rural and northern Canadians. The argument that costs would be passed on to the consumer represents no change from what currently happens. I doubt I am alone in wishing that internet access were more socialized in this country; it is the entire body of our youth who will be our future citizens, not just the children of people who can afford $60+/mo. for their high speed internet. If ISPs were required to meet higher standards of infrastructure on a national scale I would be behind them in lobbying for some federal tax benefits or other consideration.

What I do not want to see more of is those with more money being able to surf at higher speeds than those without, as is currently the case. Additionally there ought to be rules that require ISPs to accurately represent their actual speed and not just a theoretical maximum of data transmission. ISPs should never be allowed to enter into more contracts than can be honoured with their current technology/infrastructure, and certainly not to then curtail my speed in response to traffic or anything else.

[updated 2009-04-20 15:32]

Any traffic management system that attempts to limit some users over others or charge heavy users more is wrong for the simple fact that it doesn't cost an ISP more to provide the service. Moreover, the cost of upgrading networks is becoming cheeper and cheeper.

Saul Hansell's article in the NY Times clearly outlines the redundancy of ISPs arguments.....

(http://www.nytimes.com/2009/04/20/business/20isp.html?_r=1&hpw)

April 20, 2009
As Costs Fall, Companies Push to Raise Internet Price

By SAUL HANSELL
Internet service providers want to end the all-you-can-eat plans and get their customers paying à la carte.

But they are having a hard time closing the buffet line.

Faced with rising consumer protest and calls from members of Congress for new regulations, Time Warner Cable backed down last week from a plan to impose new fees on heavy users of its Road Runner Internet service.

The debate over the price of Internet use is far from over. Critics say cable and phone companies are already charging far more than Internet providers in other countries. Some also wonder whether the new price plans are meant to prevent online video sites from cutting into the lucrative revenue from cable TV service.

Cable executives say the issue is not competition but cost. People who watch or download a lot of movies and TV shows use hundreds of times more Internet capacity than those who simply read e-mail and browse the Web. It is only fair, they argue, that heavy users should pay more.

“When you go to lunch with a friend, do you split the bill in half if he gets the steak and you have a salad?” Landel C. Hobbs, the chief operating officer of Time Warner Cable, asked recently in a blog post defending the company’s now abandoned plan.

Still, critics say the image of Internet providers as restaurants about to go broke serving an endless line of gluttons simply does not match the financial or technological realities of the industry.

They point out that providers’ profit margins are stable, and that investment in network equipment is generally falling.

These plans to charge for above-average Internet use “are unjustifiable for almost everywhere in the country except for rural America,” Richard F. Doherty, the research director of the Envisioneering Group, a consulting firm that studies cable technology.

Cable or telephone networks have little in common with a restaurant, the critics say, because there is no electronic equivalent of food to buy. If all Time Warner customers decided one day not to check their e-mail or download a single movie, the company’s costs would be no different than on a day when every customer was glued to the screen watching one YouTube video after another.

That is because their networks are constantly being expanded to handle ever-greater peak periods. It is the modern equivalent of how the old AT&T was said to have built the long-distance network to handle the number of calls expected on Mother’s Day.

“All of our economics are based on engineering for the peak hour,” said Tony Werner, the chief technical officer of Comcast. “Just because someone consumes more data doesn’t mean they drive more cost.”

Yet even as the providers continually upgrade their networks, the cost of the equipment needed to do so is shrinking steadily, reflecting the well-worn economics of computing.

Indeed, the equipment needed to add capacity to any household costs a fraction of one month’s Internet service bill. Comcast, the nation’s largest cable provider, has told investors that doubling the Internet capacity of a neighborhood costs an average of $6.85 a home.

The cost of providing Internet service is about to fall even more, as cable companies install new technology, called Docsis 3, that will both increase their capacity and allow them to offer much faster download speeds.

So far, however, companies in the United States have chosen to use Docsis 3 as an opportunity to offer far more expensive Internet plans. Comcast has introduced a new 50-megabit-per-second service at $139 a month, compared with its existing service that costs about $45 a month for 8 megabits per second. Time Warner just announced it will charge $99 for 50 megabits per second.

By contrast, JCom, the largest cable company in Japan, sells service as fast as 160 megabits per second for $60 a month, only $5 a month more than its slower service.

Why so cheap? JCom faces more competition from other Internet providers than companies in the United States do.

Cable systems in the United States use the same technology and have roughly the same costs. Comcast told investors that the hardware to provide 50-megabits-per-second service costs less than it had been paying for the equipment for 6 megabits per second.

Questions about the speed, availability and affordability of Internet service in the United States will be central to the study Congress has required from the Federal Communications Commission next year. And cable and phone executives are worried that the commission may call for more regulation of Internet service, which currently is free from any government price controls.

Time Warner Cable abandoned its plan to expand a test of what it called “usage-based pricing” in four cities after Senator Charles E. Schumer, Democrat of New York, announced his opposition to the idea in a meeting with Glenn A. Britt, the company’s chief executive.

[updated 2009-04-20 16:34]

The "best effort to all" approach. If the capacity is deficient, then it's time for the ISP to upgrade the service to support the demand.

The consumer does not mind paying a "reasonable" cost of service but supplying shareholders with larger and larger profits in lieu of serving the public is not acceptable.

[updated 2009-04-20 17:34]

How can you have a managed approach when Bell and Rogers have a monopoly on the last mile? If the ISP's don't want to invest in their infrastructure (and recover their costs accordingly) then perhaps they should have their monopoly taken away. I believe the "last mile" should be nationalized then leased back to Bell, Rogers, or whoever, at a fair price. There are other larger ISP's in the world who could build and support this "last mile" and probably make a good buck as well.

[updated 2009-04-20 19:47]

I believe network wide management is the best approach. Suppliers can not be allowed to control access to content by choosing services.

[updated 2009-04-20 21:35]

The best approach to traffic management in a state-sponsored monopoly such as Canada's high speed broadband access market, is "No Internet Traffic Management", please.

In a true open market, I wouldn't see a problem with access providers implementing or not these type of things. In Canada's monopoly market, this is a measure used by the monopoly agent to try and avoid competitors in the form of wholesalers or in the form of innovative p2p applications e.g. p2p TV broadbcast, direct video on deman, etc

[updated 2009-04-20 22:05]

Prohibit throttling and traffic management. Don't allow ISPs to continue.

[updated 2009-04-21 05:29]

The future of wireless needs to advance along with the rest of the world and Canada is lagging more and more every year. The days of minutes, voice plans and data "options" will come to an end once 4G gets a foothold. Where will Canada be once this happens? The Canadian Wireless establishment is trying to slow that progress and halt innovation and competition.

It’s a pretty simple concept that the wireless industry is trying to protect its own interests and continue with their present business model. If they had their way everything would stay the same and the Telco's would continue to make deals with other companies in order to use their network to advertise or sell its product with exclusive deals. The side effect of this is that innovation and competition gets stalled and the internet is now managed and manipulated as the carriers see fit to fulfill the advertising goals of its partners. The thing is the internet is not owned by a Company, it is owned by the public and should remain that way. Companies need to advance their network in a way that sustains and adapts to a neutral internet that is free of bias information and advertisment if they wish to base their network on the internet. I realize that it’s a fine balance of supply and demand for ISP's to offer data on a network, but the world is global and Canada needs to compete with similar costs and tech. So if there is one common thing that all of us should share, then this would be to allow an internet that is free from restrictions alongside more competition. If a company wants to offer caps and higher costs to balance capacity then this is fine, but governments must regulate in order for competition and inovation to flourish in Canada so Canadaian companies are kept in check and Canadians productivity keeps pace with the world. I don’t mean allow foreign ownership but to allow foreign companies to have an attractive reason to invest in Canada, similar to Vodapone and Verizon or AT&T and Rogers deal way back when. Also, to allow new companies to grow and not be swallowed up by the established ones.

With an unrestricted wireless internet alongside real competition, this will open itself to Mobile VOIP, video chat, IPTV, internet radio, real time road conditions and other forms of communications from a multitude of providers along with an unlimited amount of ip based devices which combine to increases innovation and productivity throughout Canada and constant contact with the world including more and more remote areas.

So don’t get caught up in the idea that Rogers, Telus, Bell want you to believe their network will “fall apart” and be less "secure" if data is made more available or if any device would be allowed to connect to its network to VOIP and stream ect... This would not be true if the wireless industry was identified as an ISP and regulated more closely by the CRTC in order to increase and sustain competition which in turn forces wireless to innovate and overcome todays capacity and security issues. By doing so this increases productivity and additional competition for Canadians forcing more inovation, leading to more jobs and so on..

[updated 2009-04-21 15:47]

For resources that are not shared or are not bottlenecks (think: the last mile of an ADSL connection), the provider has no business managing traffic.

For shared resources, the provider has to be a referee with respect to sharing.

Of course the referee has to be fair, but what does that mean? It certainly means that the ref cannot privilege his own family (i.e. other products that the ILEC offers). It means that that the ref should not discriminate based on race (i.e. network protocol).

The referee should share resources among the players (customers). But only to the extent that the resource is shared. If one customer's packets are crowding out another's, it is appropriate to shape the traffic to more evenly share. This is a kind of subscriber-level management, but not like the introduction outlines.

Note that an last-mile provider's customers are not just end users but other ISPs. Managing the ISPs customers is only that ISP's business. The last-mile provider might provide tools to the ISP to help the ISP manage its own customers but control must rest with the ISP.

ISPs could well allow customers to use QoS capabilities if TCP/IP to prioritize their own packets. That could allow users to self-manage real-time constraints. The RFCs don't impose a common understanding of QoS so a per-ISP regime is the best that can be hoped for, but even that could well be useful. There be dragons here.

[updated 2009-04-21 16:22]

The question should not be 'how to manage' ... but 'who' should be allowed to manage Internet Traffic.

There are a great many 'points' within all Internet sessions that are already managed, with several different objectives. Many of these objectives have nothing to do with improving the user's Internet experience but are better described as intended to destroy the business opportunities of competitors and increase the revenue opportunities for Public Telecommunication providers.

The most disruptive of these are initiated by the computers connected in between individual users and their ISPs ... traffic to be disrupted is often identified using a DPI function.

In Canada we have now managed to create a disfunctional Internet that has become unreliable and is no longer trustworthy.

This appears to be the intended result of the actions of the very few corporations that control the infrastructue used to provide Telecommunication Services to the Canadian Public.

I believe that these corporations (in particular Bell Canada) have deliberately sabotaged Internet Services within Canada in order to create a justification for their unethical (and probably illegal) imposition of a 'walled garden' to replace a viable Internet.

What Bell Canada, Rogers, and Shaw have been calling 'Network Management' and 'Throttling' is actually a massive rotating Denial of Service attack primarily directed at Independant Internet Service Providers and their customers.

If we allow this massive Denial of Service attack currently in progress to continue the negative economic consequences for Canada will be substantial.

I believe that the Goverment of Canada should prohibit providers of Public Telecommunication Services from also providing Internet Services.

We cannot have a viable Internet without first having a 'neutral' Telecommunications infrastructure.

[updated 2009-04-22 12:35]

The solution that makes the most sense to me is assigning each user a minimum guaranteed level of bandwidth, and allowing them to burst to higher levels as network utilization allows.

This allows for the maximum utilization of available resources while protecting realtime services like VoIP; and several systems (HTB, HFSC, etc.) already exist for this purpose.

[updated 2009-04-22 14:28]

The best effort approach is the only approach that makes sense because as a consumer, we are being promised 7mbps speeds. Managed approach is the lazy way out since the Telecom company doesn't have to do anything but sit on their fat butts and watch as the profits increase.

How does over provisioning not make sense in the LONG TERM?!
They are a company that has the goal of making more money. One method is the increase their number of clients. Which is a given; more and more people will subscribe to high speed internet as the technology becomes more accessible. Wouldn't they have to build larger and larger networks anyways? If they're refusing to build larger networks, then they are purposely forcing its clients to have a bad user experience. They are also forcing their network to become saturated by increasing client load but not network capacity.

The only tool the telecom should use to manage their networks should be non intrusive, non port specific, non packet specific, and non identifiable. That is reasonable and keeps their clients' privacy.

I also don't mind internet usage caps if it is reasonable. At 200gigs combined, that is more than enough for the average user. Put the rates above 200gigs and no one would complain about being usage caps.

[updated 2009-04-22 19:36]

"Network over-provisioning" is the obvious answer. As any other utlity, there needs to always be more capacity than demand, especially in the event that the service is required in an emergency. Assuming that the Telcos actually implemented this and did not try to meddle in the affairs of their customers, I'd welcome paying an additional fee, so long as our Internet access is "neutral".

[updated 2009-04-22 23:41]

The problem with letting ISP's manage the network is conflict of interest, to who's benefit are they going to manage it, the end user or their own? They say that they are delivering the best service possible, but how do we know for sure, are we getting what we pay for? I think the best approach is for the ISP's to disclose traffic data, let them show just how bad it really is, let them publish the data. Let the CRTC decide how much if any management is necessary and how it should be done. There is no question that bandwidth costs, we get that, but how much and how much should the end user have to pay. Why does there have to be a limit in the first place? Show us why. What is a reasonable limit, how do we arrive at the magic number of 60GB as a cap? Why not 30GB or 150GB? I believe that there should be no restriction imposed by the ISP at all, however there should be consequences for people who 'hog' or overuse the network, and this can be defined and for those people they should have to pay a premium.

[updated 2009-04-23 00:39]

The Best Effort to All Approach is the only sensible approach from a consumer standpoint. Why? Because it is up to consumers to decide how they want to use the service they are paying for.

Canadian companies have a bad habit of under-investing in infrastructure and then crying wolf when their under-investment causes problems. Network over-provisioning has to be the norm and perennially profitable ISPs such as Bell and Rogers that have an effective duopoly on Internet access in most of Canada should not be allowed to falsely claim poverty as an excuse to not invest in our data networks. Network over-provisioning is most definitely a sustainable long-term solution in Canada.

I hope the CRTC will introduce and enforce strict net-neutrality requirements for all ISPs.

[updated 2009-04-23 00:47]

The best effort to all approach is clearly superior from the perspective of encouraging innovation. The managed approach posits a model for internet that is backwards in the sense that it is structured around what is best for the telecoms as opposed to what is best for the Canadian economy. Though the telecoms view this issue otherwise, they are the means to an end, as most infrastructure is, not the end itself. To promote their interests over that of internet users is very much like creating an energy grid without ensuring that people can afford to run appliances using the energy grid.

Not surprisingly, from a regulatory framework, I see the CRTC's role as analogous to energy price regulators - it should approach telecom services as a public good, and the telecoms themselves as utilities. The goal is to regulate with the intention of providing the best infrastructure possible so that all other sectors of the Canadian economy can flourish.

This is not a trivial issue. I am a Canadian entrepreneur who lives in the United States and is trying to decide whether to expand my company in Canada or the United States - my option is Canada right now because the American regulatory framework is skewed towards the telecommunications companies. Internet service costs in the United States are higher than in Canada and the quality of service is lower. Caving in the the telecom position will influence my decision, and quite possibly destroy my business.

[updated 2009-04-23 08:44]

The correct approach to Internet traffic management is “best effort to all”. An even playing field is needed, and an open internet allows small independent media to grow. We need multiple sources for information, managing the internet helps the large media-corporations to control the flow of information and eliminate small media business.

[updated 2009-04-23 09:49]