Network Presence Blog

There’s been a lot of mainstream media (MSM) commentary about the NBN, much of which is negative by heavily vested media interests, that I don’t read or pay attention to.

But during the years (now) of the NBN conceptualisation and startup, various plans and intents have been outlined by the NBN, then modified, updated and removed when some of those plans or such were found to be ‘incompatible’ (to coin a term) with the general requirements of the broad “ISP Industry”.

One of these plans currently is the number of Points of Interconnect (POIs) that the NBN suggests it will create to aggregate end-user / FTTH customer traffic and connections and to “hand-over” customer data streams to ISPs/NBN RSPs.

Currently that number is over 120 POIs across the country (PDF), which has been initially assessed by the ACCC and allowed through their process in mid 2011.

I won’t comment too often publicly like this here, but I believe that that # of POIs is (initially) too high.

See the PDF linked above for all their locations, some of which are ‘next door’ suburbs and I believe that the NBN can further aggregate suburban network traffic closer to the CBDs of each State (of Australia), because frankly that’s what a lot of fibre suppliers are going to be doing anyway, causing congestion and other commercial problems getting access at each and some POI locations and Data Centres (eg: Exchanges, which have been troublesome to gain access to via Telstra for many years).

I’ve heard suggestions that there should be only 4 POIs, and I think that’s too few (by far), but might I suggest something like 2-8 POIs for each State of Australia, say 8 in the larger or higher population states (Vic, NSW and QLD), with 2 or 3 in SA, NT, ACT and Tasmania, with WA having ~6 POIs. At least 2 in each State allows for some basic redundancy/availability (eg: if one state’s POI became ‘offline’) and ~6 (per State) may give further efficiencies for network aggregation, considerations of geographic and fibre topology, etc.

Just briefly I want to ‘drill down’ on what may result (infrastructure and commercially) from too many POIs spread across cities and regional areas:

Executive Summary: A “messy layer” of fiefdoms and performance problems, just ‘above’ the NBN’s “edge” network and if the NBN is not going to further aggregate these POIs, then I suggest that it should drop (again) its current CVC charge.

Further explaining this is that numerous fibre and other Telcos will scramble to be able to connect their networks to all of (say) a city’s or state’s POIs, then on-sell that to (wanna-be) NBN RSPs and Providers of all sorts (NBN registered or not). This will (I’m sure) mean new physical fibre builds and rollouts to link POIs, causing more urban over-builds and disruption, which immho is needless, when the NBN has the network and capability to deploy the next step of urban and regional interconnects to avoid this “mess” of fibre rollouts and scrambling to aggregate POI connectivity. This kind of ‘street side mess’ has occurred to lucrative or large CBD and urban Telstra Exchanges, by just a few ISPs across each Australian city and the NBN can avoid this again. Of course, one way the NBN could avoid this again is by ‘collapsing’ their POIs to existing Telstra Exchanges, but as above, there will be demand by new and more fibre based “NBN POI Aggregators” to these (now very old) Telstra Exchange buildings and data centres, again “making a mess” in those suburbs and areas.

A better solution is for the NBN to perform this “next hop aggregation” closer to each state or city’s CBD and existing Data Centres (where there’s much more idle and available fibre network capacity and connectivity), this may mean removing the POI “Outer Metro” layer, creating a single “Metro” layer and ‘collapsing’ some of those then Metro POIs. Such could also be done with ‘internal bearers’ between some large Regional POIs too..

I’ll go on here to respectfully suggest that if 120+ NBN POIs is going to be retained (do be aware of lobbying or such by vested interests who already see opportunity in getting to as many relevant POIs as early as possible), then the NBN should drop its CVC price, because it’s not going to be reflective of the full cost to get end-user ISP customer data streams to CBD (like) Data Centres and ISP networks.
Finally, Telstra Wholesale offers a similar service to its “Layer 2” ISP customers in each state and the NBN could use that as an upper limit on further CVC re-pricing considerations.

So I’m suggesting then a total of 20-30, max (say) of 40-50 NBN POIs across the country, by the time that the NBN is approaching full rollout (which is in many years still).

I believe that the ~30-40 NBN POIs level may reduce some of the issues which have been quite recently commented on by large ISP owners during their business sale announcements and which to my calculation would cause substantial Infrastructure and cost “overruns” to any ‘growing-to-national’ NBN RSP operation and business.