Can you clarify regarding “fair queues”? Would the customer with 100 TCP active TCP connections get a larger share of the available capacity than the customer with 1 active TCP connection? What about the video stream or software download that opens 10 TCP connections vs the video stream that opens 1 TCP connection?
The other thing that strikes me is customers want predictability. OK, maybe less so with mobile, but certainly with fixed. So I either can or cannot stream 4K shows on Netflix, or 2 HD shows simultaneously, or DirecTV On Demand. I can or cannot Skype with my son in Afghanistan. If I could do it yesterday, but not today, I am pissed off. If I pay $5 to watch an on demand movie and the first 10 minutes goes fine but then peak hour comes and I can’t and it tells me I have to download to my DVR and watch later, I am pissed off. So a best effort approach to speed (and therefore to what applications will and won’t work) may lead to a poor customer experience. Unless you have some way of notifying customers, or letting them pay for priority. Some applications will degrade gracefully. Netflix is pretty good about adaptive stream quality. I don’t get many customer complaints about video quality, but people sure hate it if the video stops to buffer, or displays an error message that their Internet has slowed. The opposite view to this would be why limit someone to a certain speed tier when the network is not “congested”, leaving capacity unused. From: Christopher Gray Sent: Thursday, January 14, 2016 11:41 PM To: [email protected] Subject: [AFMUG] Metered Plans / Full Speed To Customer? I had a conversation with a network professional who was very interested in the idea of a "metered" plan. His thought was to open up the customer connections to full speed and run fair queues instead of throttling bandwidth. Pricing would be based on usage, but with very low rates compared to cellular or satellite (e.g., 100 GB for $60). The three main thoughts were: 1) Knowing that speeds would be better in off hours (somehow promoted or advertised) could get users to operate outside of peak times thus reducing peak load on the network. 2) Customer prices would more accurately represent their load on a system. 3) Plan sharing would not be a significant concern, as usage would rise and cost would rise. Now, I can see those benefits, but I have these specific concerns.: 1) If everything is opened fully today, network performance can only get worse over time as subscribers are added. 2) Variability in speed over the course of the day may cause customer concern. 3) Many video streaming services seem to suffer with variable bandwidth availability. Any thoughts on this method of providing service? It seems very close to the cellular plans where speed is almost never mentioned, only data use. I have some ideas to make such a service work, but I'd like to know others' thoughts and experiences. Thanks - Chris
