[OpenAFS] Re: jumbo

Marc Dionne marc.c.dionne@gmail.com
Mon, 28 Feb 2011 23:18:12 -0500

On Mon, Feb 28, 2011 at 13:32, Harald Barth <haba@kth.se> wrote:
>> (fragmented packets are actually more efficient in our RX
>> implementation than having to process more single packets)
> Are there actual numbers from this century to validate this
> theory? If yes, I would be interrested to see those.
> Harald.

Of course your mileage may vary, but for example a particular
throughput test that I use gives something like this for execution
time with -jumbo on the fileserver:

0.036u 0.568s 0:08.04 7.3%      0+0k 0+0io 0pf+0w

and without -jumbo:

0.031u 0.579s 0:23.44 2.5%      0+0k 0+0io 0pf+0w

So this test gives me almost 3x the throughput with -jumbo, even
though neither of the interfaces on the client or server are
configured for larger packets and everything goes over the wire in