A better calculation of the value of an internet exchange

Bill Norton, dr. Peering these days, has made a calculation of the value of an internet exchange . You may remember a similar article written by me a while ago. My calculation was back of an envelope type. Bill has a more scientifically sound calculation and a better list of things you need to account for. The numbers aren't too different: AMS-IX saves the industry roughly 25 million a year. Definitely worth a read and a great site to learn about peering and transit.

Me moving house fools Google Latitude

I enabled Google Latitude on my mobile phone, just to have a look at how it works. I must say it works pretty well. I had expected it to do so as the Google Maps program on my Nokia E51 was already pretty accurate in finding where I was. (600 meters is good enough in general).

Most of the time it seems to be using the GSM Cell-ID. The accuracy it provides is a good indicator. GSM is between 600 and 1500 meters, sometimes less. However, at home it's different. At home, in my house, the accuracy is off by 100 kilometers. The location it gives is the location of my old home in Rotterdam. I now live in Almere as of two months. I figured out that anywhere around the house where my wifi access point isn't visible, that's where I am in Almere, but not when my access point is visible. :-)

So if ever Google Latitude tells you you are in Rotterdam, while you are certain you are in Almere, that's when your within 30 meters of my house

Mobile telco's fail against Reding on mobile termination

Quite a bit on Mobile Termination fees in the news today . In the EU big Mobileco’s made a last ditch effort to persuade ms. Reding and mr. Barosso not to cut rates. In India new entrants asked the government to abolish mobile termination rates all together. India is a great example for the EU, on the effects of cutting mobile termination rates. In the EU the average mobile termination rate is 8.7 euro cent (and as high as 15.09 cent in Bulgaria) In India there is a proposal on the table to bring down mobile termination from 0.48 euro cent a minute to 0.16 euro cent per minute (or 1% of what the Bulgarians pay). Small operators of mobile networks even call for the full abolishment of termination charges. Meanwhile mobile usage in India has increased from '326 minutes per subscriber per month in October 2003 to 464 at present'. That's 2 times more than in the EU!

BTW the mobileco's are scaremongering when they say that their revenues will drop. What is most likely to happen when termination rates drop in the EU, is that the net outpayment from fixed lines to mobile lines will drop dramatically, saving users of fixed lines (old people and companies) a bundle. The ARPU's of mobileco's don't need to drop down dramatically as they can offer more minutes of use in bundles that cost about the same as current bundles. In the US and Singapore mobile termination fees don't exist and ARPU's are comparable or above those of the EU. ($29-46 in Singapore and $52 for T-mobile USA). Minutes of use in both countries is way higher than in the EU.

Now we'll have to wait until the negotiators of the T-mobile Germany, Vodafone UK, Telefonica and Orange France have frustrated the discussion in the European Council with the Commission enough

What Free (France) could do with a mobile licence

Chances are increasing that Free (the maverick DSL/Fibre Telco in France) could lay its hands on 5mhz of mobile spectrum (and maybe some more). Free's CEO Xavier Niel has promissed that he will drastically change the game for mobile calls and save the french consumer €1000 a year.

The French mobile incumbents must be worried as Free has changed the game in the French DSL market with its 30 euro pays all offer (and all literally means anything you can think of). Here are some ideas of what Free could/should offer that we are currently not seeing. (And yes you could also see this as my wishlist for mobile telco's)

1. No data charges (30 euro pays for all data and calls to fixed and to international numbers)
2. No SMS charges (He isn't charging for e-mails now is he?)
3. Integration of fixed and mobile calling. Who sais you can't call a fixed number to reach a mobile number. 
   
4. Integration of mobiles with online chat systems (MSN, Google chat, AIM etc)
5. Integration of mobiles with Free's harddisk recorders for television. Hook your mobile up with the HDD-recorder.USB, Wifi, 3G don't care :-)
6. Full integration of fixed and mobile networks: Piconets, DSL backhaul, FON-like Wifi access (hey Xavier, why don't you have FON-like wifi-access yet). If you're a customer, you can connect to my network
   
7. Cheap pre pay SIMs for foreign roamers. Niel started with providing cheap calls :-) Cheap roaming is dependent upon the operator of the foreigner roaming in France and requires the EU to get a clue.

Something that won't change is mobile termination fees. Unless the regulator fixes it. High costs for calling to mobile will remain in France. It's such a nice free money earner :-)

Anything I've forgotten that Free, given its position on the Telecom Cool Wall, would need to implement?

The value of an internet exchange (roughly)

An internet exchange generates value for the market parties connected to it and to the wider community. The components of this value are:

1. the value of transit traffic saved for the ISP's and telco's via the exchange
   
2. the value of transit traffic saved via direct peerings, without the IX being in the middle, in the collocation facilities where the IX is present. (also known as private peerings)
3. Increase in quality of traffic exchanged through more direct connections
4. Presence of multiple transit providers generates competition on price and quality of transit connection
5. The benefits will be passed on to business and communities  and that will generate a multiplier effect.

The first one is the easiest to calculate. For instance AMS-IX is generating more than 600 Gbit/s per second every day. Transit is at  $10 per mbit/s/month according to Telegeography, but at $4 according to friends who attended the Global Peering Forum  in the Dominican Republic. The benefit of peering is between 28.8 million and 72 million each year. With AMS-IX having a 9 million euro budget, it's clear that the members of AMS-IX benefit more from it than it costs them. Euro-IX stats show that in August 2008 total traffic exchanged was at 1.765Gbit/s. The total value of internet exchanges is therefore at least 86 million a year. Amsterdam boasts the lowest transit prices in the world, therefore the benefits easily exceed that. 

The second one is much harder to calculate. There are no reliable estimates of how much traffic is passing through private peerings at peering point locations. LINX  tried it a couple of years ago, but they stopped. Their last guestimate was that 25% of traffic didn't pass the IX. 

Quality is even harder to calculate in monetary terms. I'm not going to put a value at it :-)

Transit prices have dropped fastest and deepest in places where internet exchanges were established and were succesful. When the NDIX started in 2001 the costs of some connections dropped by 50% straight from the start. Traffic then was still well above the 500 euro/mbit/month even at major exchanges. 

Multiplier effects. Money not spend on transit can be used for more profitable purposes. A typical multiplier is between 2 and 3. 

With a rough calculation as the one above, the value of an internet exchange like AMS-IX is easily above 100 million. The value of Euro-IX members can be above 500 million

You're just a dumb pipe, now pay up.

Telco's have been eyeing content providers for money to pay for network upgrades and priority lanes. In the United States Wired reports that  ESPN, Disney and NFL have turned the tables. If ISP's don't pay them, their customers don't get access to ESPN360, matches and other content. The content providers have correctly concluded that they have the power. It is their content that people want to access. 

The NFL has demonstrated in recent years that broadcasting of NFL content is nothing more than a small margin game for cable companies. Cable companies need the content, but that also means they (or their customers) will need to pay whatever the NFL asks. The cable company is just a dump pipe and billing platform and gets paid accordingly. (The debate in the US goes much wider, and also sees anti-trust regulation come into play) 

Public broadcasters in Europe, like the BBC, omroep.nl and the Norwegian broadcaster probably can't follow this route, but telco's shouldn't find it likely for these organisations to start paying as recent fights have shown. 

Tidbits: CMT, termination rates and net neutrality in the Netherlands

On Benoit's blog Fiberevolution the Spanish Regulator CMT reacts to mrs. Reding criticizing them for their decision on the Telefonica's fibre network. They rightly state that mrs. Reding is incorrectly accusing them of using a loophole. Ms. Reding has no power over remedies and claiming she should is an abuse of democratic checks and balances. This doesn't mean that CMT shouldn't be questioned on its decision. From what I read of it, I have to disagree with it and point them to Opta's decision on KPN-Reggefiber. But still good call for them and good of them to be blogging.

In the UK the courts have brought down the terminating charges. BT succesfully argued against the height and this got them down to to 4p per minute by 2010/2011. The existance of terminating charges is still baffling from a competition point of view, but it's good to see the European regulators bringing charges down. The Indian regulator meanwhile is complaining about termination rates too. At 0.48 euro cents a minute for a fixed to mobile or mobile to mobile call they are clearly too high. The proposal is to bring them down to 0.16cents per minute. Meanwhile mobile usage has increased from '326 minutes per subscriber per month in October 2003 to 464 at present'. That's 2 times more than in the EU!

I'm also wondering which Dutch ISP's will be found out by Google's new net neutrality labs. Every now and then I hear rumours about shaping taking place. We'll see.

UPC will not fight KPN-Reggefiber-Amsterdam deal

Everybody has written about the KPN-Reggefiber deal already. Webwereld has one new thing to add and that is that UPC will not fight the deal. At least that is what their spokesperson says.They might change their opinion of course, but it doesn't seem likely. They lost the previous round already and there is no indication the city of Amsterdam wouldn't be allowed to do this or for KPN and Reggefiber to get a too big market share.

UPC however isn't going down without a fight. Their margins are great (after the almost bankruptcy a couple of years ago), so they have some space to fight KPN. Furthermore their current offer looks very appealing. I'll take some time later to have a shootout between KPN Glasvezel and UPC.

Internet religious zealot? moi?

Simon Torrance of Telco 2 just called me an "internet religious zealot". Yup. Moi. :-) Oh well....

It was a very interesting reaction to a comment I made on the Telco 2.0 blog. I had reacted to an article and online video in which Andrew Bud of Mblox presented "sending party pays", as a new business model. His argument was that it would solve many of the problems the mobile industry has with data costs, if the sender of the data was charged instead of the user requesting the data.  My main point was that it was an interesting business model for ringtone type content. I didn't feel that it was a cure for all the problems described, mostly because of the effect of termination costs. The internet doesn't use sending party pays for this reason. It uses a model called peering and transit.

Simon went on:
... you seem to be displaying the outrage we so often see in the Internet religious zealots, who elide the supposed impracticality of the solution with the negative effect it would have on those who currently free-ride.

I'm afraid you're showing a poor understanding of the business model you are mocking. You say "Just think of the pricing madness that we would see if every network in the world would have the pleasure of setting its own prices for the incoming traffic it receives". But this is exactly how the successful, multi-billion dollar application-to-person bulk SMS business actually works!!!

To this rather harsh statement  I reacted with a second, bit more thought out, response.  Andrew Bud (the Mblox gentleman ) then  reacts to my reaction that

This is a very cogent argument indeed. What I think it neglects is the fact that value, cost and charging are beginning to move badly out of alignment in the ISP world.

This leaves me wondering whether I'm a madman or whether I've actually understood the problem well. If I look at the two statements, I think the answer lies in three elements of the response of Simon and Andrew: freeriding, costs and the overloading of networks.

What it comes down to is whether you believe that

1. content providers are using the internet for free and/or
   
2. costs of carrying traffic are too high and/or
3. that networks are getting overloaded. 
   

 If the above arguments are correct, then this might be a reason to go to the content providers and ask for them to participate more fully in the upkeep of the network. I'll try to evaluate the statements:

• The argument that the content providers are free loaders has often been heard. Most of those arguments rest on a complete lack of knowledge of peering and transit or a misinterpretation of the "free" that peering offers.  Bits don't float around the internet for free. You and Youtube pay for sending bits around the world. The prices aren't as high as when I entered this business, but an ISP/content provider who forgets to pay the transit bill is not an ISP/content provider for long. With transit down to $4 to $10 per mbit/month and lower, a typical content provider can send 100Gigabytes for 10 dollars around the world  (leaving 200 Gigabyte slack for off-peak). With the right peerings a significant portion of traffic can be exchanged via peerings saving on the transit bill. Do pay attention: transit prices (and port densities) are currently such that at its General Meeting in November last year, AMS-IX announced it would stop taking orders for new 10 and 100 Mbps ports. So at larger IX's like Amsterdam 1Gbit ports as the minimum for peering. In my opinion the argument that someone is freeloading on the net doesn't hold.
  
• The second argument is harder to verify. Are the cost of running a network such, that bits are expensive? Certainly in the UK the costs of backhaul are high and ISP's have been locked into a price war for some time. It is therefore understanble that some UK parties have the view that everything that generates traffic (iPlayer, Youtube, Bittorrent)  is bad and expensive. On the other hand, the UK is quite backwards with bandwidth speeds and is still living in the up to 8mbit/s world of ADSL. Operators in other countries are not as fearful of bits and seem to be able to run a profitable business. NTT in Japan thinks 900Gigabyte of upstream traffic is enough for their customers and has put its monthly limit there (no download limit). This is three times more than any Brittish DSL user can send from their pc. Free in France is providing up to 28mbit down for €30 (with phone, HDTV, remote access to your harddisk recorder etc) and is still making a decent profit. It seems that the costs of operating a network are different in various parts of the world and the perception of what new applications cost to the ISP vary as well. The case of ISP's in mainland Europe and Asia tells us sending more bits on existing networks shouldn't necesarrily lead to a dramatic decrease in ISP profitability.
  
•  The second item can also be viewed from the perspective of mobile networks. On mobile networks the consumers costs of 1 gigabyte of traffic can vary from anywhere between 10 euro/month for an unlimited internet connection to 1.5 million euro if you were to use SMS as the data carrier. Since technically it's all bits  we can attribute the cost difference to very cunning pricing. I'll try to make an estimate of what the costs of bits on a mobile network really. is. We've all been told that it is expensive to run a mobile telco. That is why terminating rates for voice to a mobile network are much higher than the other way round to a fixed network. In general 700% more expensive. Now if we would accept that running a mobile network is 7 times more expensive than 10 euro's per gigabyte may sound about right if we take Telenet's notoriously high overage fines of 0,23€/150Mbyte. 
• The third item, that networks are overloaded with traffic? Sorry but that just isn't true. Everybody who knows the industry sees that traffic is growing with 50% per year or less. Growth is actually decelerating. Properly run networks who upgrade their infrastructure regularly shouldn't face any problems coping with a non existant flood.
  

The conclusion: Some of the reasons given by Simon and Andrew for the Mblox sending party pays idea don't exist. The fundamentals of the telco industry don't make it too expensive to use mobile data.  It is true that the cost of mobile data are too high, however it is the telco's themselves who are charging much more than the true price of bits. Mblox has a good solution to fix that problem from the point of view of the consumers, so that buying an episode of House MD on your Mobile isn't costing more than going to iTunes on your pc ($1.99). The source of the problem lies with mobileco pricing and not with the fundamentals of the industry.

I do salute the mobilecos for keeping prices up as high as they have. It is great business even though from the perspective of consumer welfare it isn't the optimal situation!

Updated on wednesday: grammar and sentences cleaned up after some comments from readers