Re: IPv4 vs. IPv6 Ethernet Performance

On Thu, 30 Aug 2012, Norbert Aschendorff wrote:

Hi,

> I tested it using tcpdump: http://nopaste.info/9394068f54_nl.html
> The length field says for each packet 1408 bytes, so that should be OK.
>
> The Wireshark instance on the iperf server says something like "16732
> bytes on wire" for the most packets (not always with 16732 bytes, but
> most packets over 10,000) - could that be reassembled somehow?

only slowly catching up on email so... chiming in now.

I'd assume in this case the iperf "server" is linux or did Jack add
IPv6 LRO support to e1000?  Sorry, I am not up-to-date.

However, any modern peice of hardware should be able to fill the 1G
link even with software doing csums or offloading really and all our
routing table lookups.

What's the well known FreeBSD machine of a machine?

I can only imaging what's going on for you and some of the latest work
was not yet merged to head or 9;  I have 1 or 2 patches posted to net@
for review and testing though.  Sorry not immediately helpful.

/bz

-- 
Bjoern A. Zeeb                                 You have to have visions!
          Stop bit received. Insert coin for new address family.
_______________________________________________
freebsd-stable@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-stable
To unsubscribe, send any mail to "freebsd-stable-unsubscribe@freebsd.org"