On 07/14/2014 09:42 AM, George Herbert wrote:
On Jul 14, 2014, at 6:03 AM, Jared Mauch <jared@puck.nether.net> wrote:
In my experience the bandwidth is typically the lowest part of the cost equation.
Why transcode on 1k nodes when you can do it once and distribute it at lower cost, including in electricity to run the host CPU.
Centralized transcoding on dedicated hardware makes sense.
- Jared
Except perhaps for the (current discussion) small rural ISP.
The bandwidth scaling equations out in Ruralistan have never been the same as in large metros. You see this in wireless delivered performance as well. Netflix is probably not the straw that broke the camel's back, but it's The Thing Du Jour which one can point at and criticize, so it 's becoming a focal point.
Sure, but that's nothing more than the latest version of "I gambled on oversubscription as a business model, and lost." As you point out, and has been pointed out previously by several other posters, this is how the Internet works. Some new thing is always going to come along which uses more bandwidth than previous things, and if that new thing gets popular ... In Brett's case he made the point explicitly that it's not even a matter of his rural customers not being able to get service if his prices increase to cover his actual costs; it's a situation where if he raises prices he will lose his customers to his competition. (Which in all likelihood have prices for the rural customers which are in some manner "subsidized" by other customers.) So yeah, "Survival of the Fittest" sucks if you're not the fittest, but that's life sometimes. Doug