From: Djerk Geurts (djerk@djerk.nl)
Date: Tue Jul 31 2007 - 10:37:30 ART
The key question is: is the switch capable of knowing the same MAC on tro
different ports both in different vlans.
Both hosts must be in the same segment to communicate to be able to resolve
ARP addresses. As anything outside of the local segment is forwarded to the
default gateway (and off the vlan or back onto it if it does lollipop
routing. But when routing why loop the vlans? Just use a trunk interface to
the router instead or use two cables between the router and the switch...
Am I right? I'm ignoring bridging environments as they involve gateways as
well though maybe not routed they do forward the arp requests. Though I'm
not sure whether they alter the arp request to make the receipient accept
it. Thinking about it at L2 the receipient would accept it if the
destination IP adress is his. Imho the source would not broadcast for a host
on another segment. This kinda reminds me of some vague I struggled with
during my preparations for the written.
Djerk
> -----Original Message-----
> From: nobody@groupstudy.com [mailto:nobody@groupstudy.com] On
> Behalf Of Ben
> Sent: dinsdag 31 juli 2007 15:23
> To: Carlos G Mendioroz
> Cc: martijn; ccielab@groupstudy.com
> Subject: Re: Bridging VLANs by looping cable
>
> Just when i thought I had it figured. What happens at ARP level?
>
> Just to break it down.
>
> 1. Host in vlan1 sends ARP broadcast.
> 2. broadcast flooded to all vlan1 ports, including port with cable
> interconnecting vlan1 to vlan2.
> 3. afaik, vlan2 port should drop the broadcast. Now I'm not sure what
> happens at this stage.
>
> assuming we put aside our routing knowledge for a moment:
>
> 4.If broadcast flooded through vlan2, destination host receives it and
> unicasts its IP back.
> 5. Unknown unicast flooded to all vlan2 ports including
> interconnection
> port.
> 6 host in vlan1 caches entry and sends unicast data. Or does
> it? When it
> detects the source as being from another IP subnet, shouldn't
> it drop the
> arp response packet? If from same subnet, it will accept the entry.
> 7 If same subnet, indeed they should communicate. If not, I'm
> not sure what
> happens next.
>
> Perhaps someone else can demystify this communication further.
>
> Ben
>
>
> On 7/31/07, Carlos G Mendioroz <tron@huapi.ba.ar> wrote:
> >
> > Hmm, you are for a surprise if you think so.
> > Yes, this works, and there are some setups where this may
> prove useful.
> >
> > Like some switches that only support one private vlan
> promiscuous port
> > and you need to connect a couple.
> > Or some restrictions on where you can rate limit on some
> architectures.
> > (E.g. you can go to a real L3 interface from a L2 access)
> >
> > Nothing really clean, and I would not recomend doing that
> anywhere, but
> > works... like most hacks.
> >
> > -Carlos
> >
> > Ben @ 31/07/2007 07:13 -0300 dixit:
> > > Martijn,
> > >
> > > VLANs by design confine traffic to the vlan, unless it
> has been routed
> > > across to another vlan. Merely connecting a cable between
> two VLANs
> > > therefore does not circumvent this behavior, even if
> hosts in the two
> > vlans
> > > use the same IP subnet.
> > >
> > > Ben
> > >
> > >
> > > On 7/31/07, martijn <groupstudy@martijnj.nl> wrote:
> > >> Lim,
> > >>
> > >> i needed to laugh a little at first. Never tried.
> > >>
> > >> Not THE answer, I think you need to test.
> > >>
> > >> let's break up your q's.
> > >>> done before?
> > >> No.
> > >>> Can a host on VLAN359 now communicate with a host on VLAN360?
> > >> After i read your post trice, i see you want to do it @
> one switch.
> > when i
> > >> want to do comms with an host it starts with ip, if that
> is not local
> > >> (logical AND) I go for the default gateway. Do your pc's have
> > >> one?????????????
> > >>
> > >> When I look to your Q form another angle, say mac-addr
> table, (never
> > >> tested,
> > >> group?) I would say mac-addr binding goes to direct
> connected prot
> > first,
> > >> after that I assume that the link would for a L2 link
> between vlan's.
> > Any
> > >> takes?
> > >>
> > >> martijn
> > >>
> > >> ----- Original Message -----
> > >> From: "Toh Soon, Lim" <tohsoon28@gmail.com>
> > >> To: <ccielab@groupstudy.com>
> > >> Sent: Tuesday, July 31, 2007 9:43 AM
> > >> Subject: Bridging VLANs by looping cable
> > >>
> > >>
> > >>> Hi All,
> > >>>
> > >>> This may sound simple and mundane but for curiosity
> sake, I need to
> > seek
> > >>> clarification.
> > >>>
> > >>> Has anyone tried "bridging" two VLANs by looping a
> cable to the same
> > >>> switch?
> > >>> E.g. in the following diagram:
> > >>>
> > >>> Gi0/1(VLAN359)----
> > >>> |
> > >>> |
> > >>> Gi0/2(VLAN360)----
> > >>>
> > >>> The same cable connects to interfaces Gi0/1 (VLAN359) and Gi0/2
> > >> (VLAN360).
> > >>> Can a host on VLAN359 now communicate with a host on VLAN360?
> > >>>
> > >>> What's the implication doing this, e.g. from STP point
> of view? Is it
> > >>> recommended at all?
> > >>>
> > >>> Any advice is appreciated.
> > >>>
> > >>>
> > >>> Thank you.
> > >>>
> > >>> B.Rgds,
> > >>> Lim TS
> > >>>
> > >>>
> >
> ______________________________________________________________
> _________
> > >>> Subscription information may be found at:
> > >>> http://www.groupstudy.com/list/CCIELab.html
> > >>
> ______________________________________________________________
> _________
> > >> Subscription information may be found at:
> > >> http://www.groupstudy.com/list/CCIELab.html
> > >
> > >
> ______________________________________________________________
> _________
> > > Subscription information may be found at:
> > > http://www.groupstudy.com/list/CCIELab.html
> > >
> >
> > --
> > Carlos G Mendioroz <tron@huapi.ba.ar> LW7 EQI Argentina
>
> ______________________________________________________________
> _________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Sat Aug 18 2007 - 08:17:42 ART