Automatic TCP send and receive socket buffer sizing
chrcoluk at gmail.com
Thu Dec 21 04:22:20 PST 2006
On 20/12/06, Jeremie Le Hen <jeremie at le-hen.org> wrote:
> Hi Andre,
> Thank you for your work, it looks very exciting !
> On Thu, Dec 14, 2006 at 01:26:03PM +0100, Andre Oppermann wrote:
> > The
> > automatic send buffer is not perfect either and has some cases where
> > it may allocate too much resources of the host to a particular connection.
> > OTOH it does much better than the small fixed sized buffer we had before.
> This makes me think it makes easier the way to a DoS. A malicious user
> with a big pipe may open several TCP connections and then manage
> each send buffer to reach the maximum size (which is eight time bigger
> the classical one by default). This would mean it is eight time
> easier to exhaust kernel memory. In this case, how one could prevent
> his box from being a potential victim of this ?
> Thank you.
> Best regards
> Jeremie Le Hen
> < jeremie at le-hen dot org >< ttz at chchile dot org >
> freebsd-net at freebsd.org mailing list
> To unsubscribe, send any mail to "freebsd-net-unsubscribe at freebsd.org"
I think the opposite, without this patch my send window set to 256k
for 'all' connections to allow decent speeds.
With the patch most connections will be just 8k in size and some be 256k.
so worst case scenario with patch during a DOS they will all use 256k
windows but without the patch they would all use 256k regardless.
p.s. waiting still for releng 6 patch :)
More information about the freebsd-net