Shaking Up the FTTP Industry

Every once in a while I see something in the equipment market that surprises me. One of my clients recently got pricing for building a gigabit PON FTTP network from the Chinese company ZTE. The pricing is far under the market price for other brands of equipment, and it makes me wonder if this is not going to put downward price pressure on the rest of the industry.

There are two primary sets of electronics in a PON network – the OLT and ONTs. The OLT (Optical Line Terminal) is a centrally located piece of equipment that originates the laser signal headed towards customers. The OLT is basically a big bay of lasers that talk to customers. The ONT (Optical Network Terminal) is the device that sits at a customer location that has the matching laser that talks back to the OLT.

ZTE’s pricing is industry shaking. They have priced OLTs at almost a third of the price of their competition. They have been able to do this partially by improving the OLT cards that hold the lasers and each of their cards can connect to twice as many customers as other OLTs. This makes the OLT smaller and more energy efficient. But that alone cannot account for the discount and their pricing is obviously aimed at gaining a foothold in the US market.

The ONT pricing is even more striking. They offer a gigabit Ethernet-only indoor ONT for $45. That price is so low that it almost turns the ONT into a throw away item. This is a very plain ONT. It has one Ethernet port and does not have any way to connect to existing inside wiring for telephone or cable TV. It’s clearly meant to work with WiFi at the customer end to deliver all services. Their pricing is made even more affordable by the fact that they offer lower-than-normal industry prices for the software needed to activate and maintain in future years.

This pricing is going to lead companies to reexamine their planned network design. A lot of service providers still use traditional ONTs that contain multiple Ethernet ports and that also have ports for connection to both telephone copper and cable company coaxial wiring. But those ONTs are still relatively expensive and the most recent quotes I’ve seen put these between $200 and $220.

Using an Ethernet-only ONT means dumping the bandwidth into a WiFi router and using that for all services. That means having to use voice adapters to provide telephone service, similar to what’s been used by VoIP providers for years. But these days I have clients that are launching fiber networks without a voice product, and even if they want to support VoIP the adapters are relatively inexpensive. This network design also means delivering only IPTV if there is a cable product and this ONT could not be used with older analog-based cable headends.

ZTE is an interesting company. They are huge in China and are a $17 Billion company. They make a lot of cellphones, which is their primary product line. But they also make a lot of different kinds of telecom gear like this PON equipment. They claim they FTTP equipment is widely used in China and that they have more FTTP customers connected than most US-based vendors.

This blog is not a blanket endorsement of the company. They have a questionable past. They have been accused of bribery in making sales in Norway and the Philippines. They also were fined by the US Commerce Department for selling technology to North Korea and Iran, both under sanctions. And to the best of my knowledge they are just now trying to crack into the US market, which always is something to consider.

But this kind of drop in FTTP pricing has been needed. It is surprising that OLTs and ONTs from other manufacturers still basically cost the same as they did years ago. We generally expect that as electronics are mass produced that the prices will drop, but we have never seen this in a PON network. One can hope that this kind of pricing will shake up other manufacturers to sharpen their pencils. Larger fiber ISPs already get pricing cheaper than what I mentioned above on today’s equipment. But most of my clients are relatively small and they have little negotiating power with equipment vendors. I hope this shakes the industry a bit – something that’s needed if we want to deploy fiber everywhere.

A New PON Technology

ONTNow that many fiber competitors are providing gigabit Ethernet to a lot of customers we have started to stress the capability of the existing passive optical network (PON) technology. The most predominant type of PON network in place today is GPON (gigabit PON). This technology shares 2.5 gigabits of download data among up to 64 homes (although most providers put fewer customers on a PON).

My clients today tell me that their gigabit customers still don’t use much more data than other customers. I liken this to the time when the industry provided unlimited long distance to households and found out that, on the whole, those customers didn’t call a lot more than before. As long as you can’t tell a big difference in usage between a gigabit customer and a 100 Mbps customer, introducing gigabit speeds alone is not going to break a network.

But what does matter is that all customers, in aggregate, are demanding more downloads over time. Numerous studies have shown that the amount of total data demanded by an average household doubles about every three years. With that kind of exponential growth it won’t take long until almost any network will show stress. But added to the inexorable growth of data usage is a belief that, over time, customers with gigabit speeds are going to find applications that use that speed. When gigabit customers really start using gigabit capabilities the current PON technology will be quickly overstressed.

Several vendors have come out with a new PON technology that has been referred to as XGPON or NGPON1. This new technology increases the shared data stream to 10 gigabits. The primary trouble with this technology is that it is neither easily forward nor backward compatible. Upgrading to 10 gigabits means an outlay for new electronics for an only 4 times increase in bandwidth. I have a hard time recommending that a customer with GPON make a spendy upgrade for a technology that is only slightly better. It won’t take more than a decade until the exponential growth of customer demand catches up to this upgrade.

But there is another new alternative. Both Alcatel-Lucent and Huawei have come out with next generation PON technology which uses TWDM (time and wave division multiplexing) to insert multiple PONs onto the same fiber. The first generation of this technology creates four different light pathways using four different ‘colors’ of light. This is effectively the same as a 4-way node split in that it creates a separate PON for the customers assigned to a given color. Even if you had 64 customers on a PON this technology can instead provide four separate PONs of 16 customers. But with 32 customers this becomes an extremely friendly 8 customer per PON.

This new technology is being referred to as NGPON2. Probably the biggest benefit of the technology is that it doesn’t require a forced migration and upgrade to existing customers. Those customers can stay on the existing color while you migrate or add new customers to the new colors. But any existing customer that is moved onto a new PON color would need to have an upgraded ONT. The best feature of the new technology is that it provides a huge upgrade in bandwidth and can provide either 40 Gbps or 80 Gbps download per an existing PON.

This seems like a no brainer for any service provider who wants to offer gigabit as their only product. An all-gigabit network is going to create choke points in a traditional PON network, but as long as the backbone bandwidth to nodes is increased along with this upgrade it ought to handle gigabit customers seamlessly (when they actually start using their gigabit).

The big question is when does a current provider need to consider this kind of upgrade? I have numerous clients who provide 100 Mbps service on PON who are experiencing very little network contention. One strategy some of them are considering with GPON is to place gigabit customers on their own PON and limit the number of customers on each gigabit PON to a manageable number. With creative strategies like this it might be possible to keep GPON running comfortably for a long time. It’s interesting to see PON providers starting to seriously consider bandwidth management strategies. It’s something that the owners of HFC cable networks have had to do for a decade, and it seems that we are getting to the point where even fiber networks can feel stress from bandwidth growth.



The Return of Active Ethernet

WDM_FORecently I have been seeing new fiber construction for residential service favoring Active Optical Networks (AON) over Passive Optical Networks (PON). This reverses a decade-long trend where PON had clearly won in the US market. The reemergence of AON has been sparked by Google and others who have pushed the industry to be able to offer gigabit connections to homes.

When fiber was first built to residential neighborhoods both technologies had some success in the market in the US. But once Verizon chose PON to build its FiOS network, other builders took advantage of the price reductions driven by Verizon’s large deployment and there were very few new active networks built. Around the world the choice of technology varies. Europe also has largely chosen PON technology due to a success there by Alcatel-Lucent. But South Korea went with active Ethernet due to the large number of apartment buildings in the network, where AON is a better solution.

AON technology has never disappeared in the US and it’s the technology of choice when building fiber networks to serve business districts and business parts. Metro Ethernet is one form of an AON technology. In fact, most fiber networks have a mix of technologies and many networks use PON to serve residences and small businesses but use active connections to serve large data customers like schools and large businesses.

There are pros and cons to both technologies when building in residential neighborhoods. But the two most distinctive characteristics are the way that bandwidth is delivered and the configuration of the physical fiber network. Active Ethernet requires a dedicated fiber for each customer. PON can put up to 32 customers onto a single fiber, although it’s generally some smaller number in actual field deployment. This means that the physical fiber bundles in the network have to be significantly larger in an AON deployment. This won’t make much of a difference when deploying in small towns or rural areas. But in a densely-packed urban area the extra fiber pairs required by AON can cause some concerns, particularly for fiber management where fiber connections are aggregated at headends and hubs.

Bandwidth is handled very differently with the two technologies. With Active Ethernet each customer gets whatever bandwidth the network provider supplies. Active networks routinely can deliver 1 gigabit to 10 gigabits. Larger connections are also possible but can get significantly costlier. A PON network shares bandwidth between customers. The mostly widely deployed PON technology in the country is GPON which delivers a gigabit path dedicated to delivering cable TV and a separate data Ethernet path of 2.4 gigabits to share among the customers on a given PON. That is sufficient bandwidth to give everybody a 100 Mbps connection, but it’s not enough bandwidth if you want to guarantee a gigabit connection. There is a new PON today that can deliver 10 gigabits to each PON, but many network operators still see that as inferior to direct gigabit connections when shooting for a gigabit delivery.

The other trade-off is cost and there are cases where each technology can cost less. The customer electronics generally cost more with PON. A PON network of any size requires placing huts in neighborhoods used to terminate the feeder fibers and electronics. But an active network has a greater cost for fiber since there are more fiber pairs in each bundle. But in many cases these two costs somewhat cancel out and we have seen small town deployments that price out almost identically under the two different technologies.

There was a time five years ago when anybody building a residential fiber network could not even consider active Ethernet. There were very few vendors active in the space that supported US markets. PON had won the US marketplace and when building a network once simply decided between the various PON vendors. But today it’s all up for grabs again and anybody building a new fiber network needs to give strong consideration to both technologies.