Do You Understand Your Chokepoints?

Almost every network has chokepoints. A chokepoint is some place in the network that restricts data flow and that degrades the performance of the network beyond the chokepoint. In today’s environment where everybody is trying to coax more speed out of their network these chokepoints are becoming more obvious. Let me look at the chokepoints throughout the network, starting at the customer premise.

Many don’t think of the premise as a chokepoint, but if you are trying to deliver a large amount of data, then the wiring and other infrastructure at the location will be a chokepoint. We are always hearing today about gigabit networks, but there are actually very few wiring schemes available that will deliver a gigabit of data for more than a very short distance. Even category 5 and 6 cabling is only good for short runs at that speed. There is no WiFi on the market today that can operate at a gigabit. And technologies like HPNA and MOCA are not fast enough to carry a gigabit.

But the premise wiring and customer electronics can create a choke point even at slower speeds. It is a very difficult challenge to bring speeds of 100 Mbps to large premises like schools and hospitals. One can deliver fast data to the premise, but once the data is put onto wires of any kind the performance decays with distance, and generally a lot faster than you would think. I look at the recent federal announced goal of bringing a gigabit to every school in the country and I wonder how they plan to move that gigabit around the school. The answer mostly is that with today’s wiring and electronics, they won’t. They will be able to deliver a decent percentage of the gigabit to classrooms, but the chokepoint of wiring is going to eat up a lot of the bandwidth.

The next chokepoint in a network for most technologies is neighborhood nodes. Cable TV HFC networks, fiber PON networks, cellular data networks and DSL networks all rely on creating neighborhood nodes of some kind, a node being the place where the network hands off the data signal to the last mile. And these nodes are often chokepoints in the network due to what is called oversubscription. In the ideal network there would be enough bandwidth delivered so that every customer could use all of the bandwidth they have been delivered simultaneously. But very few network operators want to build that network because of the cost, and so carriers oversell bandwidth to customers.

Oversubscription is the process of bringing the same bandwidth to multiple customers since we know statistically that only a few customers in a given node will be making heavy use of that data at the same time. Effectively a network owner can sell the same bandwidth to multiple customers knowing that the vast majority of the time it will be available to whoever wants to use it.

We are all familiar with the chokepoints that occur in oversubscribed networks. Cable modem networks have been infamous for years for bogging down each evening when everybody uses the network at the same time. And we are also aware of how cell phone and other networks get clogged and unavailable in times of emergencies. These are all due to the chokepoints caused by oversubscription at the node. Oversubscription is not a bad thing when done well, but many networks end up, through success, with more customers per node than they had originally designed for.

The next chokepoint in many networks is the backbone fiber electronics that delivers bandwidth to from the hub to the nodes. Data bandwidth has grown at a very rapid pace over the last decade and it is not unusual to find backbone data feeds where today’s data usage exceeds the original design parameters. Upgrading the electronics is often costly because in some network you have to replace the electronics to all nodes in order to fix the ones that are full.

Another chokepoint in the network can be hub electronics. It’s possible to have routers and data switches that are unable to smoothly handle all of the data flow and routing needs at the peak times.

Finally, there can be a chokepoint in the data pipe that leaves a network and connects to the Internet. It is not unusual to find Internet pipes that hit capacity at peak usage times of the day which then slows down data usage for everybody on the network.

I have seen networks that have almost all of these chokepoints and I’ve seen other networks that have almost no chokepoints. Keeping a network ahead of the constantly growing demand for data usage is not cheap. But network operators have to realize that customers recognize when they are getting shortchanged and they don’t like it. The customer who wants to download a movie at 8:00 PM doesn’t care why your network is going slow because they believe they have paid you for the right to get that movie when they want it.

Should You Build a Cable TV Headend?

I still meet new businesses all of the time who are just entering the cable TV business for the first time or who are opening up remote markets from their service core. In the past it was a no brainer to build a new headend for a new market as long as that market had enough potential customers to justify the capital outlay. But I find myself hesitating today when I am asked the question of whether one should build a new headend. I don’t think the answer is an automatic yes any longer and there are a number of reasons for this.

Transport. One huge consideration is bandwidth transport. It always makes more sense to use the signal from an existing headend somewhere as long as you can get the signal there for less ongoing cost than building a new headend. The amount of bandwidth needed to transmit a full channel line-up is huge and can easily require at least 100 Mbps. The bandwidth varies a lot depending upon the specific method that is being used to send the TV signal to customers. For example, the bandwidth needed to send a lineup that has both analog and digital tiers will be larger than a lineup that is all IPTV. And the amount of bandwidth is even greater if you want to transmit video on demand.

The price of transport varies widely by location due to the availability of fiber, but overall there has been a big reduction in transport prices. The long-haul transport business has gotten very competitive and there are a host of companies that sell not only bandwidth, but also dark fiber or fiber lamdas. Also, a number of new middle-mile networks were built with federal stimulus grants and those networks, by definition, have to offer reasonably priced bandwidth. In many cases I am seeing transport as a good alternative to building a new headend, whereas a few years ago building a headend almost always looked like a lower-cost alternative.

Aggregators. You also should consider using a network aggregator. One that many of my clients use is Avail Media. Avail has aggregated a channel line-up that comes from the satellite directly in MPEG4 format, meaning that it can be taken directly from the satellite and used in an IPTV distribution network. The advantage of doing this is in the cost savings for the headend. A lot of the capital cost in a traditional headend is spent for equipment that translates TV signals from one format to another. The cost, size and power requirements for a headend drop significantly if the TV signals don’t have to be translated.

Of course, Avail and others aggregators charge a premium for getting the signal to you in the right format and you need to do the math to make sure that there is a net savings in equipment compared to their ongoing transport charges. But many of my clients have found aggregator arrangements that have saved them money.

Headend Sharing. Before I would build a new headend today I would always look around to see if there is an existing headend in the area that I could share. Generally, almost anybody except for the major cable companies would be interested in sharing a headend. Sharing a headend can help a headend owner offset the cost of running their headend while requiring very little ongoing effort after the initial connection.

There are a number of issues to consider when thinking about sharing a headend, but I have dozens of clients who have figured out ways to share. The biggest issue is the signal format. For example, it would make no sense to share an analog headend with somebody who is operating an IPTV system. The cost of translating channels from analog to digital would be almost as costly as building a new headend. There are also contractual issues with some of the programmers who make you jump through extra legal hoops before they will agree to let you transport signal from an existing headend to a different operator in a different market. But headend sharing makes a lot of sense and today, and sharing would almost always be my preference over building a new headend.

Other Issues.  There are always other considerations to consider. For example, if you share a headend or buy content from an aggregator you are still going to have to somehow insert the local must-carry networks onto your system. So you will need to a ‘mini-headend’ of some sort that lets you add your own content to the content that comes from somebody else.

Even if you share somebody else’s headend you might want to consider operating and inserting your own video-on-demand. This will cut down on the transport needed between the locations. If the market is large enough you also might want to consider inserting your own local advertising rather than inflict ads from some distant market upon your customers.