We continue to be fascinated by the presentation by Roy Gradwell, Director of Connected Real Estate Ltd, at the Telco 2.0 Digital Cities session. We think the ideas he floated deserve a much wider audience. He presented a new option for financing network build-outs, different from existing vertically integrated models (e.g. Verizon FiOS) or Muni/open models (e.g. Amsterdam's Citynet).

What interests us most is that it provides a practical framework for realising Malcolm Matson's open access vision of the future, where networks are funded and owned by long-term low-risk investors and any service provider can ride on top. This is called an OPLAN (Open Public Local Access Network), and implies both the end-user access and metro backhaul are part of the same open network. It's an intellectually attractive proposition. The trouble is finding the route from "here" to "there".

Some of the biggest problems with municipal fibre deployments are down to the fact that it's a big, expensive, monolithic project. The up-front cost is hefty, and its repayment means you have to be very sure there will be enough demand to pay it back. It's difficult to trial the idea of muni-fibre (or any other kind of metro-fibre rollout) without making a huge investment and therefore taking a big risk. This is the "anchor tenant" problem Dave Hughes, Director of BT's Wireless Broadband division, mentioned during the session. Other speakers noted how hard it was to co-ordinate the purchase of connectivity across multiple public services given their varying contract commitments and buying cycles.

Plus, if you're the city government, you can run into problems in the courts - in some places you might get sued by an incumbent telco, and in the European Union quite a few cities have run into trouble with the legislation on state aid to industry.

On the other hand, as Roy points out, for enterprise and government users the bottleneck is between the LAN and the WAN; and in the UK, there's been hardly any metropolitan area network investment since the end of the cable boom in 1996.

The principle doesn't need too much stretching to cover residential users either - after all, there's not much difference between a LAN-wired office block, a LAN-wired factory, or a LAN-wired block of flats from this point of view, and getting fibre reasonably close to the home is the precondition of fibre-to-the-X, VDSL, WiMAX, and the like.

Nobody wants to build a metro backhaul network without access network customers; but nobody wants to build an access network without a plentiful supply of cheap metro backhaul. And few are willing to risk doing both. So, what is to be done?

Gradwell's suggestion is simple, as all the best ideas are in retrospect. There are six core elements:

• An incremental approach to building an open-access muni metro network.
• Shift from opex to capex - own, don't rent.
• A multi-utility approach to cabling and ducting around town.
• A hybrid public/private approach to access networks that connect users to the trunk.
• A stable infrastructure investment plan to encourage private actors to invest.
• A holistic view of end-user costs that looks beyond just out-of-pocket ISP and TV/phone service fees.

The incremental OPLAN

He suggests that cities, or for that matter other actors, start by taking an inventory the sites they already own and their major internal networking customers. Then link up heavy traffic generators that currently use third-party connectivity. Stop shipping your internal phone, e-mail, database synchronisation and video surveillance out from the LAN (where you almost certainly use at least 100Mbits/s Ethernet and probably Gigabit Ethernet) to a telco or ISP network (at some much lower capacity and higher price) and then back in to the LAN in another building. Rather than paying for every bit, why not run the GigE direct between the switches in the two buildings?

So step one is to look for clusters of public buildings (e.g. fire + police) that share a connectivity need and join them together.

Shift from opex to capex

With a small capital investment, these buildings can now share a common backhaul to the Internet or outside world, rather than having to buy one each from a telco. You get an immediate drop in opex from both their internal communications as well as external traffic. The business case is easy to make, particularly where the capex is low because you're digging up the street for other reasons anyway.

Multi-utility approach

It's a sort of mixture of opportunism and foresight - opportunism in that you pick out links that happen to be needed, ducts, culverts, and the like you happen to own, and add more fibre than you need, and foresight in that you deliberately seek to add fibre whenever you dig up the road, and plan to add a node when you start a new building. By managing the physical ducting as part of a multi-utility plan, you can greatly lower your costs. Replacing the pavement? Lay a fibre. Lamp posts wiring getting old? Lay a fibre. Sewers a bit too Victorian? Lay a fibre. Gas pipes looking leaky? Lay a fibre.

Today's telecoms industry seems to be an unnatural joining of infrastructure, operations and customer relationship businesses. Who would be surprised to see them go their separate ways?

Hybrid public/private approach

And then, of course, why not hook up some more buildings? But you can do better than that; if you have all this capacity, you could link up to the company across the street and charge them for it. Then you might extend it to their other site across town. And private players may decide to hook in and build access links between businesses and residential users and the backbone, and start offering retail services.

A stable plan

By publishing a long-term plan of what "open links" are going to be deployed and when, private players can start to make investment plans to piggy-back on this network. A corporation might not be around in 20 years' time, markets change from week to week, private investment plans get cancelled easily. A town or city is there permanently, and human geography changes only slowly by comparison.

Holistic view of cost

With this model for incremental deployment, you keep rolling until you cover the whole city. It has the advantage that there are no leaps of faith; you simply install links where you have a need for serious bandwidth, or where the cost of telco transit hurts. The EU and incumbents can't complain because you originally built it purely for your own needs -- right? The open access tariff is a secondary motivator.

It's only in the later stages of deployment, where the network has to add leaves to reach odd patches of city property in residential areas, that any of the usual problems emerge. But by then, you've already wired up everywhere that's reasonably traffic-dense, and that's a start. It's not restricted to city authorities, either; a group of businesses or other organisations could get started building a shared fibre network in the same way, rather like this very cool L2 Ethernet operator in Wellington, New Zealand.

By the end, you've taken a lot of cost out of public service delivery and private enterprise within the city. So by shifting their tax dollars from opex spend to capex, with a small and temporary investment blip (with quick payback), you end up with cheaper goods and services. The real cost to users isn't the telco bill they see in person; it's the ones they pay by proxy.