John Linton
Exetel has completed the activation of a 4 terabyte caching service aimed at quadrupling the speed at which an end user can download a P2P sourced file while at the same time halving the cost to Exetel of providing the file. This is achieved by using equipment and software sourced from a company called PeerApp. It has taken some two months of gradual implementation, with the expected amount of ‘teething problems’ to put this function in to production but this service is expected to be of significant advantage to Exetel users who use P2P applications to source popular files.
This is the second stage of an 18 month to 24 month implementation program to speed the delivery of P2P traffic while reducing the cost of providing that traffic that started a year ago with the introduction of a P2P bandwidth control system from Allot and will culminate in mid 2008 to late 2008 with the full implementation of an Akamai server ‘farm’ and a 12 terabyte mirror service for those files not provided via the Akamai or PeerApp servers.
This has been a very ambitious (not to mention vey expensive) project and a fair bit still remains to be done but I'm really pleased at how we have been able to foresee a major problem some two years ago and then methodiclly go about finding the very best soution to it when, at that time, there weren't any and, as US ISPs didn't have the same degree of this problem, there were no quick 'find and copy' solutions.
Our initial first step of deploying the Allot Net Enforcer caused us a fair degree of pain in lost business and 'angry' P2P customers but without it Exetel wouldn't have survived and it would not have been possible to so effectively then deploy the PeerApp solution (which didn't exist when we first went looking for it). That was an excellent, and very encouraging, start to the long and difficult process of dealing with the continually escalating use of P2P software to 'max out' delivery bandwidth. We realised that a some future time (still not here thank goodness) streaming video would totally transform the use of delivery bandwidth and we would need an effective way of doing that as the 'old delivery models' simply wouldn't be financially feasible.
We would expect to activate the Akamai servers within the next 4 - 8 weeks and then we will 'mop up' the remaining heavy downloads via a 12 terabyte mirror service.
By mid 2008 we should have reduced the cost of our international IP to less than $A200 per mbps and should be delivering close to a gigabit per second of end user files from the combined P2P and mirror caching solutions.
If we do in fact accomplish all of those objectives we will have achieved something really significant for Exetel's customers (much faster file deivery) and for Exetel (much lower cost of file delivery) and will have allowed Exetel to continue to offer the lowest cost broadband services with, by far, the highest download allowances.
To quote George Peppard from that antedeluvian TV show - "I just love it when a plan comes together".
The amount of bandwidth used by an 'average' ADSL user has almost quadrupled over the past three years - partly due to the use of P2P but more, and more increasingly, due to the use of video both 'live' from sites (like utube/myspace) and for future use from a myriad of legal and ilegal sites.
These steps taken by Exetel have only barely managed to keep the costs of providing broadband services under control and every indication is that the amount of bandwidth used per 'average' user wil continue to climb more rapidly than over the past three years.
It makes me wonder how the current bandwidth models in place at Telstra, Optus, aapt and iPrimus are going to cope as an increasing number of their customers use ADSL2 type speeds.