John Linton
....is they never stop needing to grow; even 'small' networks.
We have one of our US hardware suppliers in Sydney this week to install the new, much larger, version of the caching engines we have been using for over two years now - not because we aren't competent to install large disk arrays ourselves but because the new software is very different to the versions we have been using and all engineers, no matter how experienced and competent always benefit from significant time spent with a supplier's engineers. Hopefully the new hardware and software will be installed by Christmas and we will be able to more accurately gauge what difference it makes to the speed and capacity of the Sydney switching centres. Currently the switching capability in Sydney is around 4 gigabits/second (over 3 gbps direct and another 1 gbps split between Akamai and local peering) and we are hoping the new caching will increase, in raw terms from the current approximately 80 kbps per user to around an average of 100 kbps per NSW user if the new caching engines meet their targets. Back when we started in January 2004 we were deploying approximately 25 kbps per user so the increase is likely to be around 4 times greater than that original capacity.
We will continue to refine the new ADSL1, ADSL2 and Wireless broadband offerings for January 1st and Steve and I discussed how to complete the offerings yesterday in two discussions (one in the office and one in the 'privacy' of a local restaurant). Exetel remains a very small company but we have gradually built up a quite large network capability and over the last two years we have slowly changed the topology of the network from a 'star' to a 'fully de-centralised' topology with a much larger 'per user' bandwidth and routing power capability than we originally thought we could possibly afford some six years ago. If the new caches prove effective then we will deploy them in Brisbane and Melbourne and smaller versions in the other four locations over the first few months of 2010 lifting the capacity of the Australia wide network to close to 8 gbps by the end of June 2010...a fairly large growth from the 20 mbps we deployed when we activated our first broadband customer in mid February 2004.
So our meetings yesterday were to make some 'final' decisions on whether we were now in a position to move to 'unlimited' plans for those of our customers who may be interested in such offerings so that we could simplify our 'options' to one 'light user' plan and one 'family user' option per ADSL1 speed and per ADSL2 option type (Including phone rental, phone with another supplier and no phone/naked). It is a very hard call especially in the current 'turbulent' pricing times we expect to see from now through the end of this financial year. We decided to wait to see what the actual results of the trial were going to deliver but if they were within our expectations then there would be no reason not to offer a 'family' unlimited plan to our current users who fitted the 'family usage' profile....whatever that is when we get around to defining it.
I have little doubt that many ISPs already use various versions of caching to improve their cost of delivering data and already make the savings that are capable of being achieved. Some do it seamlessly so it is never noticed but there have been more than a few occasions in the past where a surprising number of ISPs have 'bungled' the implementation and negatively affected their customers. We don't see caching as being a major contributor to our 'bandwidth mix' but if it can add to the current Akamai and local peering then it does change the overall cost of providing the end user service by around 12 - 15% which is a very big saving. Anyway - based on our previous experiments (which we ended some months ago) we expect the new, improved versions to mke a significant impact on both speed and cost.
So, assuming that the new caching will deliver another gigabit or so of 'bandwidth' at a cost of something like $A30.00 per mbps and we can continue to spread customer usage across the full 24 hours as we have now been able to do for over 12 months we believe that it is possible to provide 'family' users with unlimited download plans which would be something that we would see as being appealing to the family 'market place. If we are able to continue to drive down the cost of 'IP' to the levels we believe are possible over the coming year then we will be able to deliver an 'unlimited' ADSL broadband service at a relatively very low cost - to family type users. A lot of ifs at the moment but we will replace the 'ifs' with actual figures once we test out the new caching and get a confirmation quote on lower cost 'direct' IP from more than one supplier.
I think the pricing 'turmoil' in wire line broad band pricing will continue throughout the coming months as those ISPs whose 'market share' is being eroded by the actions of Telstra and other more aggressive providers re-act with their 'reactions' and so the 'circle of death' will continue - whether it stops at a sensible point will remain to be seen.