John Linton
A major expense in providing an internet service is the connectivity between the ISP and the 'rest of the world' generally referred to as IP or ingress/egress traffic. Back at the dawn of providing internet services in Australia, or at least the time I first became involved in the early 1990s, there were very few options. As I recall you could either rent a 1 mbps line to a US carrier on the West Coast of the USA for around $US800,000 a year or, sometime after 1995 you could buy bandwidth services from Telstra for around 33 cents per megabyte (and, no, that isn't a typo).
Prices started to 'dramatically' fall from that time to now but only five years ago a 'reasonably good' price for IP bandwidth from a reputable carrier was still above $A400 per mbps and it only fell below $300 mbps (at least at the sub 1 gbps quantity) some two plus years ago. Earlier this year reputable carrier IP bandwidth fell again to around $250 mbps with SX excess capacity coming on line in the recent past allowing the price per mbps to fall to or below $A200 per mbps from more than one carrier.
PIPE's initiative of 'laying their own cable' and thus breaking the SX and AJC 'monopolies' appears to be becoming a reality with deliverable services perhaps becoming available early in 2009 with pricing well below that of the carriers who use the SX and AJC cables. It's a very bold move by PIPE and Exetel, as a small ISP, would be delighted to see it become both a reality and a way of reducing Pacific Ocean transit to a more comfortable pricing level.
I think there's little doubt that Pipe's initiative, although it's a long way away in terms of commercial impact, has been at least one of the 'triggers' in making IP pricing on the established cables more 'flexible' than I have ever seen them in the time I've been involved in buying IP capabilities.
P2P filtering from Allot, Akamai caching for many software distributions titles and, if we ever get it working, PeerApp caching for P2P generally has greatly reduced Exetel's reliance on 'raw' bandwidth but, today, we still spend over $400,000 a month on IP bandwidth and plan to increase that 'raw' capacity in the future at a similar rate to that which we have done in the past. (we have gone from 10 mbps to 1.6 gbps in 46 months - an average a little less than an additional 40 mbps per month for every month we've been in existence.
We have been talking to our current and possible IP bandwidth providers for some months now as all our IP bandwidth contracts end in the first half of 2008 and you don't make changes to IP providers casually nor do you do it, should you ever decide that you must do that, without a lot of thought and very careful planning. We have seen a reluctance on the part of our current providers to change their pricing downwards because they don't want to lose the 'captive' revenue they've enjoyed to date while making life very, very hard for Exetel.
So, with a few days to go before 2007 becomes a 'commercial memory' we have got at least two viable options in acceptable contractual form that would, if we decided to go with one or other of these alternate providers reduce our current IP costs by an average of $A80.00 per mbps per month - which equates to a monthly saving of well over $A100,000 a month which is a very substantial reduction in our monthly variable expenses.
Such a saving would more than double our monthly profit, assuming all other things remained as they are today, and would dramatically change the options available to Exetel going forward in most areas of our business and how we operate our business.
It will be interesting to see what 'final' offers for IP bandwidth become over the next 6 months or so and even more interesting to see what they become should Pipe make its cable connectivity freely available at whatever pricing gives them a good return on their 'risk'.