Hi,
Thank you.
If someone have time to check this on Xen or reproduce the same situation on
KVM it will be a nice thing ;)
Yes you are wright, the flow does the all way. And response time is better
....
Sorry a little mistake , new version of the ascii art :
WAN
|
|
access port vlan 2 - connected to a physical switch
|
Openvswitch have a dedicated eth for this.
[x86 -
brcentral]
/ \
GRE Tunnel ===> / \
trunk=[0] /
\
/ \
[hyperV - br0]
[hyperV - br0]
multiple if /
\
multiple vlan
/ \
/
\
/
\
/ \
[VM FW]
[VM WEB]
Thank you for the time :)
Regards,
2011/4/15 Jesse Gross <[email protected]>
> On Thu, Apr 14, 2011 at 1:01 AM, Benoit ML <[email protected]> wrote:
> > Hi,
> >
> > Same bridge but different vlan. Thank for your answer.
> >
> > Well I've done some others tests with interesting result.
>
> Hmm, that is an interesting result. It's possible, though it seems
> unlikely, that some of the netfilter information is propagated between
> the VMs and causing problems. I have more experience with Xen than
> KVM and I would be very surprised to find this, however, KVM does have
> tighter coupling between guest and host so it's possible that
> something is passing through. To answer your original question
> though, no, I don't know of anyone that has tested this.
>
> Just to clarify, in the second setup you moved the firewall to the
> other hypervisor and the two vlans are trunked over the GRE tunnel,
> right? So all traffic flows first to the firewall hypervisor over
> GRE, back to the hub, and then back to the web hypervisor over GRE
> again? You drew a VM Web on both hypervisors, so I just wanted check.
>
_______________________________________________
discuss mailing list
[email protected]
http://openvswitch.org/mailman/listinfo/discuss
