The outspoken chief executive of ISP Exetel has rejected claims that higher priced ISPs like Internode will be able to provide a better quality of network performance on the emerging National Broadband Network infrastructure, maintaining that all ISPs on the NBN will be offering an “identical” experience.
Last week, both Exetel and Internode released their initial pricing plans for commercial services on the NBN. As with current ADSL broadband pricing, Internode’s prices were substantially higher in a number of areas, leading some in the industry to speculate that the extra costs would be sunk into traditional areas of strength for Internode — such as network performance and customer service.
However, in a post on his blog this morning, Exetel CEO John Linton rejected the idea that the networks operated by different ISPs could perform differently on the NBN fibre infrastructure being rolled out around Australia.
“I already see various ‘claims’ being made about ‘how much better’ one ISP’s NBN Co service will be than another,” he wrote. “I am assuming people who make such statements are lunatics whose comprehension of both reality and communications networks is far removed from the actuality of the NBN Co network.”
“My knowledge of network engineering is far from comprehensive but it doesn’t have to be to understand that NO ISP can influence, in any way at all, the ‘performance’ of an NBN Co service from the customer’s preference to the hand of point between the NBN Co network and the ISP network in, currenty, a capital city data centre. Every and user customer connected to the NBN Co network will have the identical experience (whatever that may be).”
Linton said it was the same situation for ISPs who bought wholesale services from existing telcos like Telstra, Optus and, AAPT. If the wholesaler had enough backbone bandwidth to serve customers on their networks, he said, then there would be no network congestion in effect. However, Linton added, if there was congestion, then it would affect all customers — “irrespective of ISP”.
“Presumably even the lunatics and dummies can understand that simple piece of networking ‘lore’,” he said.
Linton claimed that all an ISP could do to negatively or positively affect their performance on the NBN was to under-provision the connectivity virtual circuit component (capacity to users) of the NBN service provided, or the overarching IP bandwidth made available to the NBN Co services. However, Linton added, if an ISP did under-provision bandwidth, it would be obvious, and no customer would sign up for a service with the ISP concerned.
Internode managing director Simon Hackett has severely criticised NBN Co over the past few months for what he sees as crippling bandwidth charges that will make it difficult for ISPs to offer services at a reasonable price. After releasing Internode’s first commercial NBN pricing last week, the executive published a detailed blog post on the subject.
Linton said the speculation about the quality of service which ISPs would provide under the NBN was linked to Hackett’s public comments.
“I assume this particular line in lunacy has commenced because Internode published very high NBN Co prices (which will never be used in real life) as part of their political campaign against CVC charges and requirements,” he wrote, noting the only issues that he expected to determine NBN end user pricing were “real commercial ones” which wouldn’t be in place until the nation’s largest telcos Telstra and Optus worked out their own models for wholesaling NBN fibre and selling services to their own customers.
“The subsequent hysteria in the media shocked even me who never expects any knowledge or even common sense from the overwhelming majority of ‘journalists’ covering the communications industry,” wrote Linton. Internode has been invited to respond to Linton’s comments.
Image credit: Delimiter
Linton can be confusing at times.
He’s right in saying that under or over provisioning of customer services to the CVC (the contention ratio) will affect the performance, but he can’t really say that Exetel will have the same performance as Internode without knowing what their policy is when it comes to contention ratio.
Internode may so the highest they want is 20:1 for consumer, meanwhile Exetel may say 50:1 (I’m pulling these numbers out of thin air by the way so don’t quote this as fact), quite obviously the overall performance of the former would be better than the later.
Really it all comes down to the old saying, you get what you pay for.
This is spot on. How much CVC an ISP buys has a very direct influence on the “experience” that people will get.
Look at what happened in the very early trial in Tasmania – one of the ISPs – (I think it was even Internode? or iiNet?) – didn’t originally have enough upstream bandwidth available out of the area, and people were getting far less than their provisioned speed. As soon as the bandwidth was upped, the problem was solved.
The amount of CVC an ISP buys will have the exact same ability to affect the user experience. It’s not “bandwidth” as such, just a choking point on the network if they don’t buy enough.
John clearly needs to do some basic network engineering training, but in the meantime I’m happy to lend him some of my expertise. One ISP can easily have a performance edge over another by having their leased links between PoPs properly designed and provisioned. Cut-price ISPs like TPG play their game by oversubscribing leased links and either implementing heavy QoS or simply hoping that their network traffic is spread-out enough that nobody notices. Depending on what market segment you’re targeting, this strategy can either be very effective or severely detrimental to your business, or any point in between for that matter. It will be this, among other non-last-mile-related issues that will be the differentiating factors in a post-NBN market in Australia.
If you like pretty pictures (like me), Internode are nice enough to provide a high-level overview of their leased links between their various domestic and international PoPs (their ‘backbone’ in engineering parlance), which can be found at:
http://www.internode.on.net/pdf/network/internode-international-ip-network.pdf
http://www.internode.on.net/pdf/network/internode-domestic-ip-network.pdf
These documents are very much a benchmark of how an Australian ISP backbone should be designed, but are hardly indicative of how every Aussie ISP sets themselves up. Perhaps John would be interested in making Exetel’s backbone structure available for comparison?
Exetel do put out a network map (http://www.exetel.com.au/files/exetel_network.pdf) and actually provide customers with over two dozen MRTG graphs to monitor various links.
Impressive. I wasn’t aware of this, thanks for the heads up.
What’s interesting in the Exetel map is the difference in bandwidth between domestic routes. Compared to Internode you can easily see where congestion could happen. You do get what you pay for
Except the MRTG graphs which Col mentioned (which are mostly only available to Exetel customers) show that all the links are nowhere near a point where congestion can occur. Plus Exetel has a policy of regularly upgrading those links to prevent any possibility of congestion occurring.
Except that MRTG graphs are an indication of the average utilisation, not peak utilisation. Thus, a link can be saturated for short periods of time but actually appear as though there is headroom when looked at as an average.
Wholesale pricing has been publicly provided by NBN. It pretty clear from that exactly what allowances ISPs are making for CVC etc.
I can say with a pretty high degree of confidence that the network engineers at Exetel would have access to more detailed information than the basic MRTG graphs they place on their public website.
I don’t expect to pay a premium on how well an ISP can pass data to the NBN, I pay a premium on how well they can collect it from the rest of the internet.
lol very nicely put :)
*I assume this particular line in lunacy has commenced because Internode published very high NBN Co prices (which will never be used in real life) as part of their political campaign against CVC charges and requirements,” he wrote, noting*
*the only issues that he expected to determine NBN end user pricing were “real commercial ones” which wouldn’t be in place until the nation’s largest telcos Telstra and Optus worked out their own models for wholesaling NBN fibre and selling services to their own customers.*
1/ sounds like Exetel is suggesting that Internode will end up accessing the NBN via a wholesale aggregator.
2/ given that NBN end user pricing is determined by “real commercial” factors which aren’t in place b/c Telstra and Optus haven’t finalised their wholesale pricing models,… Internode and Exetel’s published NBN retail pricing is just one big guess?
Oh dear… this all looks worryingly like COMPETITION. What a travesty.
I was going to make the same comment. Many ISP’s have always wished they could compete in a level playing field. But be careful what you wish for. If you have nothing to distinguish yourself from other vanilla ISP’s on the NBN then why would a customer choose you over another?
My knowledge of network engineering is far from comprehensive…
Says it all really!
The fact that Linton is commenting on this at all confirms to me he knows that different ISPs will have different performance.
If there are 121 POIs and CVC costs $20000 Gbps there has to be a difference as there is a substantial variable that can be “tuned” that will have a big impact on the costs of operation for an ISP.
Don’t forget the CVC is only from the POI to a local exchange, so you need multiple CVCs per POI.
Keep adding those costs up :)
If you are happy to pay $20, $30 more for the same product then more power to you.
“If you are happy to pay $20, $30 more for the same product then more power to you.”
That’s the point though, it isn’t the same product. The tail into a customer premise may be the same, but how that then is fed into the ISP and what the ISP does with the data will mean each “product” will be described in the same way but may behave differently. A 100:1 CVC contention will mean the product will behave differently at times to a 250:1 CVC contention (which is the contention NBN are basing all their modelling off).
What will be confusing for customers is that they will have no visibility of this (as they don’t currently with Exchange backhaul and AGVC contention with ADSL providers) and so can’t make an informed choice.
Seems most people are also ignoring the fact that not all ISPs have the same International connectivity and same peering arrangements both locally and Internationally. These things have a great effect on the end performance of an ISPs offering, whether on the NBN or on ADSL.
I think the NBN will shake up the industry in more ways than one. It will really come down to the actual links to the internet/peers that an ISP has as the NBN has previously indicated that they will aggressively target ISPs that show substandard performance (most likely those that don’t buy enough cvc bandwidth). So on a NBN level the field may well now be equal, but certainly on a end user service level.
Im curious how sustainable *un-metered* 40Mbps **upload** for $50 per month will be….
….I can see many companies getting this for offsite backup purposes, running the 100s of gigs through it….which, Im sure will throw a spanner in the works…..
Linton, I am not sure what to make of your comments. Sometimes quite brilliant, sometimes scatty. This is the latter.
All and very well and good Linton then you start terminating accounts as deemed unprofitable.
Networking is just plain old boring logistics.
It doesn’t matter whether you’re delivering envelopes, passengers, cold meat, or data.
Logistics is logistics.
For much the same reasons as some freight companies have competitive advantages, so too can NBN resellers have competitive advangages.
Why do people in Australia chose FedEx when ordering from Amazon?
It’s all ultimately delivered by Australia post, so what’s the difference?
Why do we have trains, planes, trucks and ships? Wouldn’t one single method work best?
The answers lie in coordination, efficiency and service.
It’s more than just getting your goods.
How long did the delivery take?
Were the goods, or their boxes damaged in transit?
Was anything lost?
Was the customer service suitable to your needs?
What tracking options do they have?
Obviously, as a customer you don’t much care about FedEx’s contracts with shipping and air-freight companies. You don’t care about Australia Post’s relationship with these same companies. You don’t need to care how good the working relationship is between these parties, nor about their financial relationships. Nor do you care how they label their boxes for more efficient transport.
What you care about is getting the grade of service delivery at the price you’re willing to pay.
This forms the basis of the companies reputation.
A good reputation suggests that these aforementioned agreements are sound and well coordinated, which allows that company to charge a premium for the same delivery.
A poor reputation suggests that these agreements may not be as sound, the relationships might be frayed, or that the company does not sort their goods as effectively.
What we’re going to see under the NBN is what we see every time we recieve a parcel, step onto a train, catch a plane, ride a taxi, bus, tram, make a phone call, or buy fruit in the middle of the city.
We’re going to see a host of different grades of service that all appear at face value to achieve the same goal.
They’re going to achieve it in very different ways, and with different results.
A lot of the time it’s not going to matter, or be appreciably different. But anyone who’s ever waited an hour in an airport for a delayed plane has probably seriously reconsidered their choice of airline.
John Linton is glossing over this fact.
As Cadbury said, “It’s not what you do, it’s the way that you do it”
Quite true, but if you look sideways to another example NRMA’s contracts with repairers stymies the whole repair and insurance industry. We really don’t get the best deal even though our insurance prices might seem to come down.
If the NBN does likewise, then the whole of Australia.Net will be a pain. and that is essentially what every body is saying.
One question to delimiter is can you through inputs here design a great alternative plan ?
Actually if anyone remembers 2006 when Exetel claims that “Internode’s capacity alleged to be less than Exetel’s” In whingepool.
That was pretty funny.
Got to love the Exetel network maps.
I feel sorry for people in Hobart, Perth, and Adelaide who want to access servers in Sydney. 50Mbps!! You have to be kidding yourself. You will have one user connect to the NBN at 100Mbps and they can’t get that speed because Exetel have limited access between those POPS.
Ha good luck with your prices Exetel. No chance
Maybe I am missing something, but isn’t Linton arguing the same thing that you commenter’s are all calling him out on, that the link from the customer to the POP will not be a bottleneck, but the connectivity beyond that??
What I read him as saying is that once it hits the ISP’s network. IE; the bit owned by Linton, and not the NBN Co, is where performance will change.
This is the point at which contention raises its ugly head. Hence, no ISP can influence the performance of a customers transfer performance UNTIL its in their network. This is the differentiator. I dont see him claiming anything otherwise.
My two cents, bigger ISP’s will obtain cheaper upstream bandwidth and result in cheaper customer pricing and smaller ISP’s will offer boutique services and top levels of support, but have a higher price tag associated with their services.
Business will be a big winner, with high speed links dropping dramatically in price.
Local content will spring up, and I cant wait for the revolution in this regard. If I were a small media content producer, I would be taking my energies into working out the upcoming distribution opportunities.
I don’t agree with Linton.
Let’s just ignore contention ratios which I am sure are far worse on TPG/Exetel than Internode.
I am on TPG at the moment in WA, all of my data flows via Sydney and onwards to either Japan or USA on TPG’s network, or into AAPT and onward to the other nearby places (eg NZ, PNG, FIJI). For the connectivity mentioned here – it is my opinion there is not much difference between ISP’s.
However, remember I live in WA, not NSW, as hard as it may be for Linton and the general population of Australia to understand not everyone lives in the eastern states ;)
Internode have capacity from Perth to Singapore, and then onwards to most of Asia – very few ISP’s do this, although *some* do have connectivity from Perth to Singapore. Exetel are not one of them.
Now comes the part that makes Internode the front runner for me, Internode rent capacity with TATA between Singapore and London, this capacity travels through India and then the Middle East, this means people in Perth can get to Europe quicker than people in Sydney as they are not hopping all through USA and the rest of Australia.
Internode are the only ISP that do this that I am aware of.
So, for people in WA, possibly SA, accessing European content – Internode will offer you a superior service from the others.
Just doing a comparison on the Internode LG, 300ms on my TPG connection vs 400ms on Internode. The data takes the most efficient path otherwise on TPG also, so it’s not like TPG have crap connectivity (per>syd>sjc>nyc>lon)
It’s purely because noone has seen the need for Sin>Lon capacity, not even iiNet who have a massive customer base in WA.
That is a huge difference in latency, and it’s actually usually a bit less, but some sort of maintenance is going on inside TATA’s network making it take a longer route.
TPG subscriber numbers are around 600,000. Internode are what, around 200,000?
Sure International transit and peering matter for a small number of power users who are interested in latency, but I think that the fact of the matter is these small number of users are vastly outweighed by people searching for better value for money.
This will become even more the case in an NBN world where TPG/iinet/internode cannot differentiate themselves by installing DSLAMs, everyone will be buying from NBN at the same price.
What will the differentiator be then? Who knows, but whoever is willing to go in at the lowest margin when internet essentially becomes a commodity like electricity is a very good chance at picking up oodles of subs.
You could say that service will be a differentiator, but when you have fibre all the way to the home, line faults are going to be rare as hens teeth, and dealing with faults seems to be when service really matters.
Be that as it may (all very good justifications) it DOES NOT invalidate his comments. In fact, it supports them as far as I am concerned. Internode are offering you a boutique service by providing access to international resources via better links out from a POP located more closely to your physical location. What happens between you and the POP at Internode that routes (requests) traffic to your preferred host will matter little from one ISP to the next, but will matter greatly once within the ISP’s network infrastructure boundaries. As you mention, Internode can deliver your traffic to your more highly valued destinations via less hops and therefore reduced latency. This is a differentiator, but the NBN Co link, is not.
Time to find a new photo of John Linton
hahaha so true
Is this the only photo of him? I’ve never seen another. Maybe we should offer a reward for anyone that gets another shot of him in the wild. ( and no I don’t mean to shoot him in the wild, although he does look a bit like a rare panda there)
Hehe… I was going to same the exact same thing. :D
This Exetel CEO loves taking stabs at Internode. Way to cover up your own companies shortfalls.
Also that Exetel domestic network map sucks. There links between capital cities are absolute rubbish.
Anyone who saw hordes flock – because of TV advertising and price – to Dodo, TPG (since much improved) and Vodafone data (reportedly improving in patches), only to get seriously burned, will raise an eyebrow at John’s suggestion “no one will buy [dearer services]”.
To save ten bucks a month, some of those hooked by the advertising claims suffered outages, poor performance, heavy mobile data surcharges, and ultimately punitive exit fees to flee to whatever they perceived as quality (and sometimes even to BigPond).
The NBN will certainly expose providers with chronically inadequate upstream provisioning and also those with poor customer service or even billing systems that trap folks into costs they didn’t expect. I wish John Linton well, as Exetel targets the low-end bang-for-buck crowd, who can be very demanding customers indeed.
The great thing about the common universal wholesale price of the NBN is that everyone in the country will get a great last-mile experience – the best that can be economically provided to their address. And this means every customer who wants or needs an excellent upstream service to go with it will be able to buy one from a provider who is not hamstrung by the fading ADSL copper or over-stretched wireless of the past.
I have lodged a submission with the ACCC at http://www.accc.gov.au/content/index.phtml/itemId/996674/fromItemId/142 outlining the serious consequences of their artificially inflated POI requirement and petitioning them to leave the POI count decisions to NBNCo engineers.
Submissions close on 8 August if you have anything to add.
Linton’s a loony but I love the way he’s prepared to speak his mind and say thing that no one else would say. I’ve got a lot of respect for the guy, even though I can see the mile-wide holes in his argument.
“He talks complete nonsense, but he does so with conviction, and for that I respect him.”
Uh huh.
Chuckle, good old predictable John Linton.
I wonder how this spew fits into his pattern of blaming everyone else from end user to Telcos for Exetel’s bad network performance.
I think mostly the difference will be quality of helpdesk and customer support, but yeah CVC contention will probably also make a difference, and quality of upstream pipe, peering agreements, etc.
Since download direction is usually more subject to contention pressure than the upload direction, you might find some differentiation in the types of queuing algorithms ISPs can apply on the download path to achieve fair distribution at peak periods (e.g. automatically identifying and throttling file sharing, email, or other types of traffic that people don’t immediately wait for a response from… but accelerating WWW and interactive traffic so the perception is a fast link). I predict that the boxes that can do this will start to become pretty much essential to ISP’s working in an NBN style situation where making the most of your CVC payments directly delivers customer experience and profitability.
Ha really goes to show just how little little John linton does know.
the NBN will be the same as it is now… and Linton’s business will be just as crap as it is now.
As they will to high a contention ratio JUST like they do NOW, amongst other things.
Internodes network and contention ratio’s are the best in the business for a reason.
None of which John boy understands.
I can’t begin to comprehend Linton’s chain of logic.
Yes, the connection from the home via NBN fibre to the handoff point is the same regardless of your choice of ISP. Nobody EVER questioned that. What they *did* question was the relative performance of ISPs.
His argument is that no ISP would under-provision the [CVC or backhaul or international bandwidth] because if they did nobody would use the ISP (people use Dodo…), so he assumes that they must incorrectly think that certain ISPs can magically speed up the last mile, and attacks that point with full force. I don’t know how he made that leap.
Honestly, this line of his says it all:
>I am assuming people who make such statements are lunatics
Yes, when he sees something he doesn’t understand, he immediately assumes that others must be crazy while he remains sane.
Comments are closed.