John Linton
I've been working on the best ways of reducing Exetel's IP bandwidth costs moving forward and have accumulated a fair amount of current information to consider. Over the weekend I read the recent CommsDay articles and the subsequent US media comments about the possibility of Google participating in a new trans-Pacific cable:
http://www.commsday.com/node/188
A few months ago PIPE was reported as being involved in providing another trans-Pacific cable but the details of that announcement are too vague for me to be able to understand what may become available and when it may become available to small companies like Exetal.
We will have 'firmed up' the costs of buying direct from Southern Cross within the next few weeks which, at least currently, is by far the most economical way to go but has very stringent contract conditions.
Currently Exetel spend around $400,000 a month on IP bandwidth and it's the largest single expenditure we have after the $1,000,000 or so dollars we spend with Telstra Wholesale each month for ADSL1 'tail' circuits. It's therefore a very high priority in terms of the cost reductions we need to put in place over the coming two years and we have to at least make a start on reducing those costs in the immediate future.
And it's becoming clearer that that might be something to be even more wary about than has been the case in the past.
Despite the reports in the press earlier in the year that seemed to state that some ISPs were claiming that IP costs are rising that has never seemed to be the case based on the information that I've been privy to. Today the 'offer cost' of IP at around the 1 gbps level is less than $A200.00 per mbps which is around 20% lower than Exetel currently pays on contracts that have a few months to run. Generally Exetel has always had to sign 24 month contracts for IP to get a reasonable price and we've been happy enough to do that - partly because we thought the price was good and partly because we were too small to have much choice.
If PIPE and/or Google do manage to get a new trans-Pacific cable in operation by mid 2009 I can see why Southern Cross itself (and those companies that re-sell Southern Cross capacity) are apparently being so easy to negotiate with recently and how surprisingly 'low' the price of SX cable capacity has become - fear of what new cable capacity will be sold at - especially if it's Google who would not only negatively affect the growth of SX's current customers but may well reduce the amounts those current customers buy from the SX consortium and its resellers.
Exetel will start to re-sign its IP contracts in March 2008 and, not unreasonably - as in the past, significant discounts are being offered for 24 and 36 month contracts with really big discounts being offered for 48 and even 84 month contracts. However if Google DOES lay a new cable and DOES get it operational in mid 2009 then signing even a 24 month contract in March 2008 becomes a much bigger 'gamble' than it appeared to be until I read the CommsDay article.
Some time in the next two months we will begin to trial the P2P caching products we contracted for earlier this month and will begin to understand the quantum of how much trans-Pacific network capacity will be 'saved' by this concept. If it lives up to the lower estimates of its developer's claims then it will reduce Exetel's IP costs by around 25% without impacting the customers in any way. If it lives up to the more optimistic claims it could reduce trans-Pacific bandwidth by 50% - but I really doubt that scenario.
Depending on how much information becomes available about Google (or even PIPE's) intentions over the next three months and depending on how much efficiency P2P caching actually delivers the permutations of what needs to be considered in making the rapidly approaching decisions on who Exetel should buy IP from is becoming increasingly more complicated.