| reply to RayW
Re: Why? Total throughput has never been the problem. In a packet-switched network it is quite possible for every user to max out their throughput without adversely effecting the network.
The problem has always been timing. If 100 people try to access google at the same time then their throughput will be divided 100 ways and the network will be 100 times slower. If 99 people access google and 1 person downloads a video at the same time then you've got the same problem. Neither the size of the files nor the amount of allocated bandwidth matters, really. What matters is the timing.
It just so happens that placing throughput caps during peak usage hours may reduce total usage time which may reduce network congestion. However, this is a roundabout way to accomplish what it really required to keep an oversold network running smoothly, controlling the timing of packet that pass through the network.
A pay-per-byte model isn't the solution for this nor is tiered service plans, packet shaping based on protocol, or discriminatory billing. The solution point-to-point timing negotiation based on IP addresses so that every user has equal and full access to the network. This is technically difficult, but but quite possible and, if done correctly, will be transparent to the end user. By manipulating timing one might increase overall latency by a few milliseconds but would still allow every user full and equal throughput up to a point. If there are too many users even this option fails, of course. |
