[p2p-hackers] Hard question....

[coderman]

On 4/1/06, Bob Harris <bob.harris.spamcontrol at gmail.com> wrote:
> ...
> I can create hundreds of TCP (or TCP-like) flows in parallel, easily consume
> more
> than my fair share of bandwidth, and easily create congestion at the routers
> by
> closing and creating TCP connections (slow start, anyone?). Many p2p apps do
>  exactly that: open many connections to many other hosts.
>
> In fact, I'm cranky at the moment because some idiot's p2p download
> is consuming
>  all the bandwidth at my current wireless hotspot. Maybe what we need is to
> extend the TCP
> ideas from the flow level to the host-level (and either embed them deep into
> the OS
> or enforce them via traffic shaping).

traffic shaping is an excellent idea and something i encourage and use
routinely.  implementing policy at the host/endpoint level is much
better than trying to kludge it within an application (throttling TCP
sockets in userspace, etc) that has a very limited view of network
capability and status.

regarding UDP: the reliable multicast charter has done a lot of work
to couple congestion avoidance and reliable transmission for datagram
transport without the full overhead of a TCP like mechanism.

my personal preference when using UDP to many endpoints (although i
admit i've focused mostly on signalling/control channels with UDP) is
to limit overall throughput to a fixed fraction of available
bandwidth.  this way TCP and other transports can negotiate session
capacity within the remaining bandwidth.