Lafayette Offers 100Mbps Residential Tier ... And Ruminations on Bandwidth Caps

Lafayette's LUS Fiber network, after recently kicking off its ad campaign, has decided to offer 100Mbps residential connections after a number of requests from subscribers. The network previously offered a 100Mbps business service for $200 -- it seems they are now just allowing anyone to subscribe at that level and price.

As John notes at Lafayette Pro Fiber blog, this is the only tier for which residential plans come with the same price as business plans.

The other residential tiers are cheaper than their corresponding business tiers by 45-48%. Nor, according to Huval's remarks in the comments is the monthly usage cap any different—in both the residential and the commercial versions of the 100 meg package is capped at 8 terabits. (Note: that'd be about 1 terabyte of hard disk storage.) The idea behind the higher prices for businesses is that they use much more bandwidth than households—and LUS pays for its connectivity by capacity.

LUS Bandwidth Caps

This brings up something I don't think I previously noted in discussions about LUS Fiber - it comes with a monthly transfer cap. I cut the cap chart out of their user agreement [pdf] above. Remember, 8 bits to the byte. Thanks to DSL Reports for the link to the user agreement.

This raises an interesting discussion. Private cable companies typically enforce caps because their network cannot physically support many users using a lot of bandwidth simultaneously. When hundreds of users share a single connection (as with cable), a few major users can seriously impact the experiences of others.

In a FTTH network like Lafayette's, there is no real danger of one user's activities affecting another's. However, there is a danger of racking up a high bandwidth charge for LUS Fiber if many users are constantly using a lot of bandwidth. By constantly, I mean really constantly as in 24/7, not as in, I watch Netflix for 4 hours each day. By bandwidth, I mean the bandwidth LUS Fiber must purchase from other carriers so a LUS subscriber can view our site (for instance). The costs for one LUS customer to transfer files with another LUS customer in-network are negligible as the traffic never leaves LUS network.

ISPs typically pay for bandwidth to the greater Internet based on the peak usage over a month. So if a few users are constantly maxing out their 50 or 100Mbps capacity, they can push that peak up considerably, raising the costs to the ISP by high hundreds or thousands of dollars.

What this means is that if a user plans to use a consistently high volume of bandwidth (one that would trip the cap), Lafayette will want to know about it and plan for it in its calculations in the amount of bandwidth it budgets.

Here we come to another difference between a typically private sector cable company and a public utility like Lafayette. In the comments of the Independent Weekly story, the head of the utility says if you expect to use more capacity than the cap allows, "Just let us know what you are looking for and we can get you a proposal."

In economic terms, when the cost of something is free or approaches free, people use it without thought because there are no consequences to them. The connections provided by recent community fiber networks suffer from this perception. In the case of broadband connections though, there are serious consequences (particularly for a smaller scale provider like LUS) and high costs when users consistently max out their connections (via unthrottled peer-to-peer services running 24/7 in the background, for instance).

Unfortunately, the arbitrary cap is an imperfect tool for solving this problem. For instance, if I transfer 50GB of data every night between 2AM and 5AM, I'll likely run afoul of the cap but I probably won't really increase the costs of LUS Fiber because even with my usage, the ISP is unlikely to experience peak usage in the middle of the night.

Another key problems with a bandwidth cap is that today's (mythical) "broadband hog" is next year's average user. Caps that do not grow over time will unfairly throttle both users and innovation.

Ultimately, almost all residential service is based on an oversubscription model. That model does not work when a significant number of users are constantly using a significant chunk of the connection -- those who are going to constantly use a connection should really have a dedicated circuit. But that is obviously more expensive for the subscriber, often prohibitively so.

I do not know what the answer is to these conflicts, but I suspect different communities will make different choices. This should be their right so long as the network and rules governing the network are accountable to the public.