John Linton
I went in to the office yesterday to interview some applicants to fill positions required by the new technologies we are beginning to deploy and the changes we are making to the directions of the company over the coming year. The interviews were, as always, interesting in all sorts of different ways including the glimpses the interviewer obtains about the ways other technology companies operate. In some ways it's difficult to objectively view what you think you learn but an overall impression I have got from interviews over the past few weeks is that many technology companies are struggling to keep up with the one constant of this industry - constant and unrelenting change at a rapid pace.
This was re-inforced, if indeed it did need re-inforcing, when I had a walk round the office before locking up. The ten large boxes holding the caching servers still 'litter' the little available spare space we have and the final equipment is due to arrive on Monday having cleared customs last Friday. As this is a major investment for Exetel and it continues to concern me greatly, both on the technical and financial aspects of such a change, I didn't lock up and leave but went back to my PC and, yet again, reviewed all of the assumptions and projections we have made in deciding to make this large investment.
The projections, assuming they are correct, are truly astonishing in financial terms - which, of course, they had to be for us to risk so much money starting 15 months ago when we took the first step of investing $100,000 in the NetEnforcer 2040. That first investment has paid for itself every month of its operation - a truly great financial decision, operationally (albeit a blindingly obvious one). While the implementation on the NE delivered an $80,000 reduction/saving in the monthly expenditure on ingress/egress bandwidth by 'pushing' 400 mbps of peak time P2P downloads in to off peak time it came at a cost of severely annoying a portion of our customers who believed that their file downloads should happen in the shortest possible time at all times of day - something we decided was not going to happen if the cost of doing that was $80,000 a month and a 25% slow down in peak periods for file downloading was not any real inconvenience. Those that didn't see it that way went somewhere else and so we lost some customers which has to be a bad thing and it would have been better if that hadn't happened.
However......I doubt that those disgruntled user's subsequent ISP didn't either already use, or implemented shortly after they moved to them, the same or similar P2P control processes that we used
We had always planned two more steps in changing the 'model' of delivering large file downloads.
Step One - The NE allowed us to 'push 25% of our peak time downloads (400 mbps out of 2 gbps) from the 8pm to 12 midnight period to 12 midnight to 12 noon period and was successful in allowing Exetel to only pay for 1.6 gbps of activated IP to handle 2 gbps of peak time demand.
Step Two - The new P2P caching is projected to allow the 400 mbps of 'delayed P2P downloads to, via the cache, be 'instantly supplied' rather than slowed down once it's operational.
So, if these 15 month old projections turn out to be true (and the first 400 mbps did turn out to be true) then we will be able to deliver 2.4 mbps of downloads at the price we currently pay for 1.6 mbps of downloads, a saving to the business of approximately $160,000 per month with, if it all goes to plan, NO customer impact on file downloads once/if the caching works effectively. The effective cost of the 800 mbps of additional downloads will cost Exetel around $A50.00 per mbps compared to a current cost for the other 1.6 mbps of $A250.00 per mbps.
An amazing advance in technology if you're prepared to do a lot of research, take some risk, understand your customer base, have very competent and dedicated network engineers and be prepared to lose more money and more customers in the development phases than you are comfortable about.
...and believe me when I say it isn't as simple as the confines of my time allowance in writing this and your attention span allowed for extraneous musings demands.
As with all neatly presented Excel spread sheet figures you've looked at and agreed with dozens of times before it all looks very likely to me.
I hope I maintain that view once the physical installation process commences some time later this month.