incubator-cloudstack-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jason Villalta <ja...@rubixnet.com>
Subject Re: Cloudstack VLAN interfaces from bridge not physical interface.
Date Tue, 19 Feb 2013 14:03:32 GMT
That makes sense but I was think more to have virtual core routers rather
than the edge routers that Cloudstack uses.

The use case would be if I have a customer with multiple vlans with both
physical and virtual hardware.  It would be nice to quick setup a core
router using vyatta with one nic and sub interfaces to be able to route
between the vlans.  That core router could then be clustered within the
virtual environment for redundancy.

I also have a need in the pre-production system I am setting up, which has
an edge vyatta router running outside of cloudstack.  I know that is
probably a No-No, but it seems to work fine except for the way the
interfaces are currently being created.



On Tue, Feb 19, 2013 at 1:27 AM, Marcus Sorensen <shadowsor@gmail.com>wrote:

> I could point you in the right direction if you wanted to do your own
> custom thing. But I think in general it doesn't scale well to filter the
> tags at the level of your 1000 virtual routers instead of your 30 physical
> hosts. Just imagine you have even 50 vlans, each with a /20, that's a ton
> of broadcast to pass in to each VM router to drop or pass.
> On Feb 18, 2013 9:09 PM, "Jason Villalta" <jason@rubixnet.com> wrote:
>
> > I am trying to implement a different kind of network stack and I was
> > wondering if someone could point me in the direction of the source code
> > that handle vlan interface create in KVM.  What I am trying to do is have
> > the VLAN sub interfaces created from the bridge interface instead of from
> > the underlying physical interface.  I have don't this in the past with
> > native kvm without problem.  The added functionality this give you is you
> > can have a virtual machine act as a vlan router by passing the tagged
> > frames directly to it while at the same time have other virtual machine
> use
> > the vlans on their normal interfaces.  This does not currently work since
> > the physical interface is being used to create the vlan sub interfaces.
> >  This pulls the tagged frames out before they can get to the general
> > bridge.  In my case bond0.10 would identify the tagged frames before they
> > could get to the bridge.
> >
> > Bond0 -> vmbr0(tagged frames) -> Vritual Router (Won't see vlan10)
> > |
> > Bond0.10 -> vmbr1 -> normal VM
> >
> > Instead do this allow both vms to access VLAN10
> >
> > Bond0 -> vmbr0(tagged frames) -> Virtual Router (Will see vlan10)
> >                    |
> >                 vmbr0.10 -> vmbr1 -> normal VM
> >
> >
> >
> >
> > --
> > --
> > *Jason Villalta*
> > Co-founder
> > [image: Inline image 1]
> > 800.799.4407 | www.RubixTechnology.com <http://www.rubixtechnology.com/>
> >
>



-- 
-- 
*Jason Villalta*
Co-founder
[image: Inline image 1]
800.799.4407 | www.RubixTechnology.com <http://www.rubixtechnology.com/>

Mime
  • Unnamed multipart/related (inline, None, 0 bytes)
View raw message