 espaethDigital PlumberPremium,MVM
join:2001-04-21
Minneapolis, MN
kudos:2
 ·Clear Wireless
| reply to atom_galaxy
Re: Is this a good thing for the net? said by atom_galaxy :
Wouldn't this be a prime candidate to fix by properly using the TOS and priority packet fields? p2p apps should set priority to minimum and games and VoIP should set it at max and there we are, low-latency for apps that need it, in a protocol-defined well-mannered way. If applications and users could be trusted to appropriately set the priority of their traffic, then yes this would work.
Unfortunately there's too much of a "screw everyone else as long as I get mine" attitude for this to work. Just look at all of the effort that these developers are going to kludge together solutions to work around ISP prioritization. When presented with the problem of "Your application is causing issues on the network" the developers turned to find ways to stealth and hide their traffic even more. The ISPs aren't innocent in this -- this escalation in coding was largely the result of overreach in management such as the Sandvine TCP RST implementations.
A few folks have suggested that Comcast's interest in P4P (ie Pando Networks) is simply to show the FCC they are playing nice with P2P. I suspect in reality the interest is being able to inject network awareness into P2P transfers using an intelligent controller. Getting your content closer to you is a nice fluffy selling point, but it does nothing to address the concerns of congestion. The problem is right now it's up to the P2P clients to decide when the network is idle or not -- if the P4P controller could limit session setup based on network use thresholds, then there might be a better coexistence of P2P traffic on ISP networks. Of course, this requires P2P users to play nice and use the P4P controllers, which pretty much means this has a very limited chance of success. |
