At 05:57 p.m. 25/02/2002 -0800, Chris Wedgwood wrote:
Because per capita, more investment has been made in local loop. I suspect that Wellington is much more densely populated than Auckland meaning the cost of a local-loop deployment more attractive for potential returns.
Richard and or Simon may be able to comment here as this clearly relates to Citylink too.
Yes Wellington is easier. Its unique geography makes it very attractive. Lack of land made them build up and so there has been more advanced engineering here for ages. The power network in Wgtn is fantastic and we basically cloned it with fiber. Wgtn has high rise, sleep slopes, short distances, narrow streets - and trolley buses. London (UK) is very similar. Doing the layer 0 engineering is great fun in both cities. There is always a tunnel, car park, basement etc that allows non traditional cabling to work. In London it can be a bit exciting when you suddenly come across the Underground where you didn't expect it, or a train rumbles past out of the blue (and you are in a void between the track and the road above). Remember its power at up to 110KV in cables so don't say a telco can't do it. UNL and London Electricity (whatever they are now) did and still do. What I'm getting at is architectures. Traditional roll outs use telco style ring or star architectures. In the power industry you NEVER run a separate cable back to the power station. Instead a mesh architecture is used and we copied it. It is unusual for us to run fiber for more than 1km, VERY unusual. Most ccts are around 3-400m to the nearest switch. But be aware that a switch may have 8 ports in use, with 3 customers. Just like the power substations. Its interesting, we regularly get vendors showing us their lovely big kit. But we never buy it. We explain our methods and they go quiet. They go away and come back with a smaller switch with some distribution and say - wow we saved 50%. Only Foundry has made the next step and gone really small and saved another 30% (memory vague here). That means we are 80% cheaper than the traditional design. Using different layer 0 engineering (yes you have to learn this the hard way) we can get into a building for say $5k (remember short fiber runs (and we carry big drills)), our MAN cabinets are live for under $2k and remember that in a typical building (under 10 floors) we will get 25% of tenants if we are VERY lucky. That means we need 2-3 ports + feeds. So why buy a 48 port switch with a chassis, dual psu etc, when in reality 12 ports is all thats needed. We also prefer to roll a truck instead of avoiding it. After all 90%+ of issues are customer related and I haven't yet found a way to ssh into a customers brain, so a small geographic area is very economical. And those 2-3 customers typically groan at paying us $300 a month, so there isn't a budget for a big switch. Another hidden is the cost of media converters. I suspect thats why some others are using multimode for the building drop. We don't because we rent dark fibers and doing the sm/mm conversion is a pain on longer ccts. Michael you want to comment ? So geography is important, but so is architecture and its vital to have a very pragmatic approach and local knowledge is vital. Thats why we're always out walking places. And I know we (ie the industry) can do better. When suppliers can get us real cheap Gig gear (as they are) we need to look at different architectures and techniques to use them. I'm trying to get our microtrenching released as copyleft or some sort of open source. That way others can use it and we (ie the country) can get Gig or 100mbps to the home at very affordable rates. Which is also why I'm against local loop unbundling - its all past its time. Why stick dsl on it and try to prolong its use - unless you happen to own it and need to pay big dividends. sorry this is so long - many interruptions......... richard.naylor(a)citylink.co.nz - To unsubscribe from nznog, send email to majordomo(a)list.waikato.ac.nz where the body of your message reads: unsubscribe nznog