I’m fairly certain that billing systems of this scale are pretty doable and not a limiting factor and isn’t a reason to not do this.
I don’t agree backhaul providers would have to change - many people who emailed me are already paying for fixed wavelength type services - this just makes their job a little easier. MMC > On 30 Jul 2017, at 4:37 pm, Paul Brooks <pbrooks-aus...@layer10.com.au> wrote: > > Back in the day*, one of the considerations in designing the NBN was avoiding > the > costs of putting in a mother-of-all traffic monitoring and measurement system > that > would retrieve and store usage data on all the 10+million endpoints. By > keeping the > product suite a set of one-time-charge and fixed monthly values, no > measurement was > required, the billing was kept simple, and periodic counter polling and > associated > data reduction, processing, storage, and robustness of systems was avoided. > Clearly that has been superseded at least on the satellite links, where NBN > are > tracking usage anyway. > > Now you are talking about the CVC layer, not end-user, so theres a few orders > of > magnitude reduction in polling and storage required to do traffic monitoring > to > implement a p95 rating system just on CVCs - but the current system for all > its faults > has the benefits that the costs and complexity of polling, calculating and > billing on > usage are avoided, and billing disputes between RSPs and NBN are fairly > binary with > the biggest possible area of contention being what date a billable element > become > billable or not. > > Also, for most PoIs, NBN switching to a p95 model of billing would require the > backhaul network transmission providers to also switch to a p95 model to > become > effective. If the backhaul providers are selling transmission between > capital-city and > POI in fixed increments that take 6 weeks to upgrade , then the NBN's > charging model > is almost irrelevant, and the incentive to delay upgrading the backhaul > transmission > retains the same problem. > In fact, its arguable how much of the congestion is caused by CVC and how > much by > unwillingness to increase backhaul capacity to manage cost. > > (*'Back in the day' means before the CVC/AVC and commercial structure was a > twinkle in > someone's eye) > > Paul. > > > > > > On 30/07/2017 6:36 AM, Matthew Moyle-Croft wrote: >> Hi, >> So, feel free to shoot me here. But curious for response either here or >> privately. I don’t live in Australia but I still do care deeply about >> broadband in Oz having something to do with it for quite a few years. >> >> The big issue with NBN at the moment appears to be congestion related to the >> amount of CVC being purchased by ISPs. The congestion is artificial because >> CVCs are bought, as I still understand it, in fixed amounts. (ie. we haven’t >> moved on from TW ADSL wholesale arrangements). >> >> Is buying the CVC in fixed amounts the right model? What if it moved to a >> “transit” like p95 kind of model? I get that there is some level of risk >> for the ISP that CVC cost could jump up a lot, but presumably it would be >> related to customer growth at a POI. It might also allow connecting to new >> POIs where you have less customers. I can see some guardrails around this - >> ie. a maximum amount maybe per POI? >> >> At least if you were smart about this it’d allow ISPs to avoid congestion as >> customer bases are growing as more houses are onboarded. >> >> Given the nature of users I don’t think a p95 number could be gamed >> particularly, but it would ease some of the peaks a bit. >> >> Thoughts? >> >> MMC >> _______________________________________________ >> AusNOG mailing list >> AusNOG@lists.ausnog.net >> http://lists.ausnog.net/mailman/listinfo/ausnog > > _______________________________________________ AusNOG mailing list AusNOG@lists.ausnog.net http://lists.ausnog.net/mailman/listinfo/ausnog