On 02/19/2010 12:10 AM, Arends, R.R. wrote:
> In our environment (gbit to 100mbit lan) i see an negative impact on using an
> 65532 tcp window size..
> Atleast with some workstations. For example with an vm in vmware workstation
> on windows i see alot of TCP Retransmissions...
>
> This is the case with 16k, 32k and 64k.
>
> Switching back to 8k or 4k shows an much smoother tcp flow.
>
> An dynamic tcp window would probably work alot better for us atleast.
>
> -reni
>
It would be useful to know if this is because of out-of-order packets,
or if it is yet another example of how utterly awful VMware's PXE
implementation is (is this using undionly/gpxelinux or using a native
gPXE driver)?
-hpa
_______________________________________________
gPXE mailing list
[email protected]
http://etherboot.org/mailman/listinfo/gpxe
