There are occasions where home-run cabling isn't a realistic option - for example where overhead cables are mandated or if the most viable option is through sewers and culverts. Such environments mean the only viable cabling system is a hierarchical one, and in the world of fibre-based NGA this means PON. The traditional view is:

• Ethernet = Open
• PON = closed

But is this always true?

The first challenge in creating an open PON network, then, is to create a network which successfully separates the passive elements - the fibre and splitters - from the active PON kit, creating the potential for multiple service wholsalers and infrastructure level competition. While the granularity of competition at this level is limited to the number of premises at the end of a splitter, it is possible to permit functional separation of the passive and active elements providing there is a very rigid design standard which creates a very clear product for the active wholesalers – no variations in the split ratios for example. In this world there is no reason why it isn't possible to have more than one wholesaler so long as they take a whole splitters worth of customers as a minimum – a good reason to keep the split ratios small.

The impact of this is that it may be possible to have multiple active providers on a single passive architecture, or at least the threat is there during contract renewal negotiations, and multiple PON standards can co-exist in any given PoP avoiding today’s world where ADSL is the only dictated standard. Infrastructure competition at the same level as LLU today, if somewhat more granular at the splitter level rather than per customer, may help to maintain levels of innovation and choice; it may be  a compromise and not perfect but it demonstrates that its possible. Sadly this has rarely, if ever happened in the current clutch of PON networks which are almost always owned and run by incumbent operators.

A second goal is to provide a cabling platform to the wholesaler which is as invisible as possible - one which doesn't constrain the services the wholesaler is able offers to the retail market. In today's market this means a passive PON infrastructure which can offer a service to the wholesaler which is as close to the performance of a home-run Ethernet network – i.e. 100 Mbps symmetrical and uncontended. Again this is possible if on fixed 12:1 split which on today’s GPON delivers 207 Mbps downstream and 103 Mbps up with no contention on the physical network. This means a wholesaler installing a PON active network will be able to offer their retail customers locating in the PoP a guaranteed uncontended 100 Mbps symmetrical service to any customer, with a further 100 Mbps downstream which could be reserved for broadcast services, for example. This is important for several reasons:

• It allows retail customers ultimate control over their contention and package differentiation – with complete control in their own network it allows them to offer 100:1 light user services through to 1:1 real-time services with an SLA, and anything in between. A higher split ratio will restrict the richness and diversity of ISP offerings. If the ISP chooses to install an E3 34 Mbps connection for internet transit, it’s their choice – but they do have a choice and customers will have the opportunity to chose between them a more expensive service based on a 1,000+ Mbps PoP backhaul.
• Secondly its symmetrical so it’s possible to treat customers as customers and not as consumers. This is critical when offering business services – a typical incumbent may opt for a higher split simply to restrict packages to asymmetrical service offering so it doesn’t raid their lucrative portfolio of higher bandwidth, high-cost business services. But symmetry is about respecting customers choice and not about control. A physical infrastructure with a high split ratio forces service types, removes choice, and limits customers to being consumers of provided services.

Given the choice, a home-run network will always be better than a hierarchical network as it allows active upgrades, competing standards, and wholesaling on a customer by customer basis, while PON is restricted to a complete splitter’s worth.  However, where the environment dictates it is possible to create a PON network with many of the same traits - in fact today's GPON can be made to outperform a 100Mbps Ethernet system by delivering a broadcast channel on the same network – but it won’t always be like that as gigabit Ethernet costs fall.

For the sake of a simily, PON is like a lethal weapon – it’s not in itself intrinsically bad but in the wrong hands it can be. It is possible to create a reasonably ethical and honest platform using PON – it’s just not done often and it still requires some level of compromise. It has all the characteristics which help an incumbent behave badly but it can also be used to create a pragmatic design within real world restrictions.