....was once the major cost of providing an internet service to a residential, or a business, user.
Back in the day (1995) when internet was delivered to residential and 'business' users over 28.8 kbps modems the cost of a 2 mbps link to the USA was around $US900,000 per mbps and comapanies like OzeMail charged up to $A10.00 per megabyte for downloads. Today current offers vary between $A15.00 and $A20.00 per mbps - one of the greatest improvements in industrial history - even if you remove the incredible price gouging applied to international telecommunications services throughout the past 40 years. Add to that the fact that even a company of Exetel's size delivers over one third of total end user IP though caching (Google, Akamai, Various peering points) the cost of delivering data to an end user is now lower than it has ever been.
It would be much lower than it is today if it weren't for the cost of 'back haul' - the cost paid by wholesale customers of the large carriers to deliver traffic from the end user to the ISP's carrier interconnect point....that cost varies between twice the cost of the IP traffic to four times the cost depending on which carrier is providing the 'back haul'. By far the highest cost of delivering traffic to a residential end user is the carrier's back haul charges followed by their 'port' charges. Many ISPs have addressed this problem by building their own DSLAM network which, courtesy of the ACCC, have allowed back haul and port costs to be reduced which, allied to the dramatic fall in IP costs should have allowed much lower prices for residential ADSL services to be offered than has proven to be the case....it isn't the case because of all the obvious reasons (Telstra less 10% remains the pricing mantra for the ISP industry in Australia).
Exetel will shortly make a decision on IP suppliers for the coming 12 month period starting April 1st 2012. We expect our 'pure' IP costs will fall around 50% from what they are today....we are already receiving the benefits of a continually growing percentage of IP delivery via caching which is saving us what would have been the previous costs of over 4 gbps of 'pure' IP bought from IP wholesalers. We will replace the current 'pure' IP bandwidth with 8 gbps of 'new' band width which will generate considerable monthly savings for us. However the port costs and back haul costs, particularly for our Telstra services will remain unchanged and will need to be subject to a review of whether we can continue to offer those services at their current costs going forward.
We have delayed, and then delayed again, more than once, making any decision on continuing to do business with Telstra while we tried to get some grip on when an 'NBN2' service might offer a viable replacement. Apart from the fact that 'NBN2' port costs and backhaul prices are even more expensive than Telstra's prices we were taking the optimistic view that they would at least be the same to every wholesale buyer (from Telstra's to Exetel's 'size'). That was always pretty naive, I admit, but it was the premise on which the 'NBN2' was promised. What is a Juliar promise worth? Nothing - but then I am as naive as the average person and did want to believe that particualar promise while always conceding it would not eventuate.
So, with some sort of real coverage delivery by NBNCo way in the undefined far future the decision on whether to continue a relationship with Telstra remains 'on the table'. We started a trial program yesterday to see if we could accelerate the change in 'balance' between our Optus and AAPT residential customers and our Telstra customers - we have never differentiated before - and over the next 2 - 3 months we will see if it is in fact possible to rapidly increase our Optus and AAPT residential ADSL customers as a percentage of our total residential ADSL customers. If this can be done then we could continue to 'hang on' to the Telstra relationship until the 'NBN2' becomes a viable alternative - or at least for another year while some definition is available on the future of the 'NBN2'. In the mean time the cost of back haul and port costs from Telstra make it very, very hard for a company lilke Exetel to continue to offer Telstra wholesale services and current negotiations simply, yet again, underline how difficult it is (for a company of Exetel's size) to deal with such a company.
Is Exetel considering setting up PoPs overseas and buying raw capacity on SXC or similar this time around?
I ask out of curiosity really. At what scale does buying layer 2 capacity, setting up a PoP (or 2) in the US/Japan and using Telstra/Optus/Verizon/AAPT for domestic transit only start to make sense (if ever)?
For each of the past three years we have looked at costing from SX direct.
The recent 44% reduction from SX direct makes it very tempting as can see our current need for 8 gbps rapidly increasing as our corporate business continues to accelerate (in the event that continues).
Right now, our needs for local connectivity (more than our peering/caching can deliver) and our need for multiple State delivery plus path diversity for back up reasons makes it still a better deal to buy from non SX direct sources.
It is a much closer decision today than it was last year.
May I suggest using your answer to my question as part of your marketing/sales materials?
As a point of comparison, if Telstra have a power failure at Paddington, they lose their access to the Endeavour cable - your architecture would not have that problem.
To meet those reliability requirements whilst having your international transit by-and-large terminating in Sydney, the amount of time you'd spend checking that paths interstate really are diverse and getting carriers to be honest about who they do fibre swaps with is a job I know I wouldn't personally want!
Whilst I understand Exetel's wish to get as many current Telstra-based customers moved off Telstra to other wholesaler-based plans, please don't forget those of us connected to RIMs and other ADSL2 blockers, who have no choice, and must remain on telstra infrastructure.