The Looming Backhaul Crisis

I look forward a few years and I think we are headed towards a backhaul crisis. Demand for bandwidth is exploding and we are developing last-mile technologies to deliver the needed bandwidth, but we are largely ignoring the backhaul network needed to feed customer demand. I foresee two kinds of backhaul becoming a big issue in the next few years.

First is intercity backhaul. I’ve read several predictions that we are already using most of the available bandwidth on the fibers that connect major cities and the major internet POPs. It’s not hard to understand why. Most of the fiber between major cities was built in the late 1990s or even earlier, and much of that construction was funded by the telecom craze of the 90s where huge money was dumped into the sector.

But there has been very little new fiber construction on major routes since then, and I don’t see any carriers with business plans to build more fiber. You’d think that we could get a lot more bandwidth out of the existing fiber routes by upgrading the electronics on those fiber, but that’s not the long-haul fiber network operates. Almost all of the fiber pairs on existing routes have been leased out to various entities for their own private uses. The reality is that nobody really ‘owns’ these fiber routes since the routes are full of carriers that each have a long-term contract to use a few of the fibers. As long as any of these entities has enough bandwidth for their own network purposes they are not going to sink the big money into upgrading to terabit lasers, which are still very expensive.

Underlying that is a problem that nobody wants to talk about. Many of those fibers are aging and deteriorating. Over time fiber runs into problems and gets opaque. This can come from having too many splices in the fiber, or from accumulated microscopic damage from stress during fiber construction or due to temperature fluctuations. Fiber technology has improved tremendously since the 1990s – contractors are more aware of how to handle fiber during the construction period and the glass itself has improved significantly through improvements by the manufacturers.

But older fiber routes are slowly getting into physical trouble. Fibers go bad or lose capacity over time. This is readily apparent when looking at smaller markets. I was helping a client look at fibers going to Harrisburg, PA and the fiber routes into the city are all old and built in the early 90s and are experiencing regular outages. I’m not pointing out Harrisburg as a unique case, because the same is true for a huge number of secondary communities.

We are going to see a second backhaul shortage that is related to the intercity bandwidth shortage. All of the big carriers are talking about building fiber-to-the-home and 5G networks that are capable of delivering gigabit speeds to customers. But nobody is talking about how to get the bandwidth to these neighborhoods. You are not going to be able to feed hundreds of 5G fixed wireless transmitters using the existing bandwidth that is available in most places.

Today the cellular companies are paying a lot of money to get gigabit pipes to the big cell towers. Most recent contracts include the ability for these connections to burst to 5 or 10 gigabits. Getting these connections is already a challenge. Picture multiplying that demand by hundreds and thousands of new cell sites. To use the earlier example of Harrisburg, PA – picture somebody trying to build a 100-node 5G network there, each with gigabit connections to customers. This kind of network might initially work with a 10 gigabit backhaul connection, but as bandwidth demand keeps growing (doubling every three years), it won’t take long until this 5G networks will need multiple 10 gigabit connections, up to perhaps 100 gigabits.

Today’s backhaul network is not ready to supply this kind of bandwidth. You could build all of the fiber you want locally in Harrisburg to feed the 5G nodes, but that won’t make any difference if you can’t feed that whole network with sufficient bandwidth to get back to an Internet POP.

Perhaps a few carriers will step up and build the needed backhaul network. But I don’t see that multi-billion dollar per year investment listed in anybody’s business plans today – all I hear about are plans to rush to capture the residential market with 5G. Even if carriers step up and bolster the major intercity routes (and somebody probably will), that is only a tiny portion of the backhaul network that stretches to all of the Harrisburg markets in the country.

The whole backhaul network is already getting swamped due the continued geometric growth of broadband demand. Local networks and backhaul networks that were vigorous just a few years ago can get overwhelmed by a continuous doubling of traffic volume. If you look at any one portion of our existing backhaul network you can already see the stress today, and that stress will turn into backhaul bottlenecks in the near future.