On Thursday 25 September 2008 at 11:35 am, Ricardo Kleemann penned
about "Re: [vbox-users] guest network performance"

> Hello Pablo,

Hi Ricardo

> Thanks for the feedback.

YW.

> Actually no, this is not on windows, it's between linux host <->
> linux guest,

Got it.

> The way I'm perceiving the performance is whenever I untar a tarball
> on the guest within the NFS-mounted directory, I see that the untar
> operation is significantly slower than what I would expect for a
> local network.

Okay, I re-read your original post; sorry for the earlier incorrect
suggestion.

As for your initial question on how to test the `host <-> guest
network performance', what I'd do is conduct several tests.

First, I'd remove `tar' from the equation.  Create a 20MB file and
place it on the host and on the guest's local drive.

I'd collect timings copying the file from the host to the NFS-mount.
Do three runs and use `time' 

Next, do the same in the VM.

How much difference is there?

The quesetion in my mind is are we running out of CPU to drive the I/O
subsystem?  Eliminating `tar' from the equation helps to understand
the problem.

I'd also be curious as to what `strace' says for a small-ish
transfer; comparing the results of each.

Cheers,
-- 
Pablo Sanchez - Blueoak Database Engineering, Inc
Ph:    819.459.1926      Fax:   760.860.5225 (US)


_______________________________________________
vbox-users mailing list
[email protected]
http://vbox.innotek.de/mailman/listinfo/vbox-users

Reply via email to