On Sun, Aug 06, 2006 at 06:27:35PM +1000, Herbert Xu wrote: > Edgar E. Iglesias <[EMAIL PROTECTED]> wrote: > > > > It makes sense, but I see a problem with this. It will remove some of the > > control applications have on how much of their data gets queued on local > > queues. Small SNDBUF's are sometimes used to somewhat avoid/control latency. > > For UDP yes but does this really make sense for TCP? >
I don't know :) The flow may end up with SNDBUF worth of skbs on the socket transmit_queue + cwnd worth of skbs on the qdisc queue + full tx ring. In slow start we'll increase cwnd and won't se packet loss until the qdisc is full (which might be at 1000 pkts). If the flow is locally limited, won't we end up with a very latent flow? where moste packets will be on the qdisc. I agree that in general this is an effect of users having to large qdisc queues, but today maybe it is beeing somewhat avoided by apps setting small SNDBUF's. So I kind of agree with what you suggest but there might be risks. Best regards -- Programmer Edgar E. Iglesias <[EMAIL PROTECTED]> 46.46.272.1946 - To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html