From: shiran guez (shiranp3@gmail.com)
Date: Sun Jan 06 2008 - 06:26:50 ARST
Hi Joseph
the traffic is not going via the virtual link it is through the virtual-link
path, there is a difference between the two. the virtual link is not a real
interface if you put a sniffer you will not see the traffic encapsulated
to the virtual link, you will see it going in the path that the virtual link
is taking.
when you use GRE then a packet is encapsulated to the tunnel and in the path
you see GRE traffic so it is different the the Virtual Link.
On Jan 6, 2008 9:36 AM, Joseph Saad <joseph.samir.saad@gmail.com> wrote:
> unless you use, no capability transit, this is true.
>
> if you have no capability transit, the traffic will be routed via the
> virtual link itself.
>
>
> http://www.cisco.com/univercd/cc/td/doc/product/software/ios124/124cr/hirp_r/rte_osph.htm#wp999437
>
> Joseph.
>
>
> On Jan 6, 2008 11:09 AM, Scott Vermillion < scott_ccie_list@it-ag.com>
> wrote:
>
> > Hey Shiran,
> >
> > I did some battle with VLs probably a few months back. While I couldn't
> >
> > offer up the link offhand, IIRC, Cisco states that the reason you can't
> > punch a VL through a stub area is because payload traffic is not
> > actually
> > sent "through" the VL; rather, it is routed natively through the
> > non-backbone area supporting the VL. Thus, all routers must have full
> > routing knowledge for the OSPF domain. Is that what you were basically
> > saying or do you have a different take on this?
> >
> > Regards,
> >
> > Scott
> >
> > -----Original Message-----
> > From: nobody@groupstudy.com [mailto:nobody@groupstudy.com ] On Behalf Of
> > shiran guez
> > Sent: Saturday, January 05, 2008 11:45 PM
> > To: paul.cosgrove@heanet.ie
> > Cc: Luan Nguyen; ccielab@groupstudy.com
> > Subject: Re: OSPF virtual link and cost
> >
> > if you see a cost of -1 then it is more then likely a bug, you can see a
> > higher accumulative distance of the links in your OSPF domain but this
> > router will not be used as a transit
> >
> > So if you have the following senarion
> > _
> > / R3
> > R5--------R1*-------*R2--------<
> > \_ R4
> >
> > the red link between R1 and R2 is set with the highest cost and you
> > advertise all the links to the OSPF, *you will not see* the link between
> > R1
> > to R5 on router's R2 R3 and R4 only in there database as the routers
> > will
> > know that a router that advertise a link with 65535 cost do not want to
> > be a
> > transit and will never will be even if you had a redundant link between
> > R1
> > and R2 and you wanted to set one of them as backup then setting it to
> > 65535
> > will not work as it will never become transit, you can set it only to a
> > high
> > metric but not to the highest.
> >
> > so with that to say OSPF Area 0 by definition is a transit area cant be
> > a
> > stub area, so by setting links in (Virtual links) trough a stub you are
> > violating the protocol definition.
> >
> > another note to that: this is the reason why you cant put a virtual link
> > trough a stub area only tunnel interface that you advertise the tunnel
> > to
> > area 0.
> >
> >
> >
> > On Jan 6, 2008 2:43 AM, Paul Cosgrove <paul.cosgrove@heanet.ie> wrote:
> >
> > >
> > > I'm seeing slightly different behaviour myself. Virtual links seem to
> > > have a maximum metric of 65535, which is also the maximum cost allowed
> > for
> > > an individual OSPF interface (so V.L.'s have the same maximum cost
> > limit
> > > as a direct connection).
> > >
> > > OSPF routes consisting of multiple links can accumulate a higher cost,
> > and
> > > a virtual link which transits multiple links can be assigned a
> > different
> > > cost depending on the version of IOS:
> > >
> > > Topology is as follows:
> > > R2----R3----R4
> > >
> > > R2 runs 12.4(5b)
> > > R3 runs 12.2(15)T7
> > > R4 runs 12.2(16)T7
> > >
> > > The behaviour I'm seeing can be summaried as follows:
> > > - If one of the routers terminating the virtual link has an outgoing
> > > (directly connected) interface cost of 65535 the V.L. goes down.
> > > - If the outgoing interface costs of the terminating routers are each
> > > individually lower than 65535 the link stays up.
> > > - Transit routers can have interface cost of 65535 without bringing
> > the
> > > virtual link down.
> > > - The router running 12.2 uses a maximum metric of 65535 for the
> > virtual
> > > link, even if the total cost of all transit links is higher.
> > > - The router running 12.4 uses a maximum metric of 65535 for the
> > virtual
> > > link, but the total metric rolls back to -1 when it increases past
> > 65535.
> > >
> > > As an example,
> > > - If outgoing interfaces from R2 towards R4 have costs of 65534 and
> > 65535,
> > > R2 shows a cost of 65533 for the virtual link.
> > > - If outgoing interfaces from R4 towards R2 have costs of 65534 and
> > 64, R4
> > > shows a cost of 65535 for the virtual link.
> > >
> > > Paul.
> > >
> > >
> > >
> > >
> > > > cost 65535 is the highest cost of OSPF if you set all your
> > interfaces to
> > > > cost of 65535 it is like you are doing stub router in EIGRP, there
> > is a
> > > > command for that if you want:
> > > >
> > > > max-metric router-lsa ?
> > > > external-lsa Override external-lsa metric with max-metric value
> > > > include-stub Set maximum metric for stub links in router-LSAs
> > > > on-startup Set maximum metric temporarily after reboot
> > > > summary-lsa Override summary-lsa metric with max-metric value
> > > >
> > > > the Virtual Link is not an actual logical interface like tunnel
> > there is
> > > > no
> > > > traffic going over the virtual link it is used to bridge the
> > backbone
> > > area
> > > > (transit area) of OSPF so if you have a stub router you cant have
> > also
> > > > use
> > > > it to transit traffic.
> > > >
> > > >
> > > >
> > > > On Jan 5, 2008 10:44 PM, Luan Nguyen <luan.m.nguyen@gmail.com>
> > wrote:
> > > >
> > > >> Do you guys know that virtual link depends on cost? If the cost
> > > =>65535
> > > >> then it will die?
> > > >> If you set the auto-cost reference-bandwidth to say...100000, then
> > most
> > > >> likely, virtual link over serial T1 links will die.
> > > >> If you have a requirement to change the auto-
> > > >> 100000,
> > > >> how do you keep the virtual link up?
> > > >>
> > > >> thanks.
> > > >>
> > > >> -lmn
> > > >>
> > > >>
> > _______________________________________________________________________
> > > >> Subscription information may be found at:
> > > >> http://www.groupstudy.com/list/CCIELab.html
> > > >>
> > > >
> > > >
> > > >
> > > > --
> > > > Shiran Guez
> > > > MCSE CCNP NCE1
> > > > http://cciep3.blogspot.com
> > > > http://www.linkedin.com/in/cciep3
> > > >
> > > >
> > _______________________________________________________________________
> > > > Subscription information may be found at:
> > > > http://www.groupstudy.com/list/CCIELab.html
> > > >
> > > >
> > >
> > >
> > >
> >
> >
> > --
> > Shiran Guez
> > MCSE CCNP NCE1
> > http://cciep3.blogspot.com
> > http://www.linkedin.com/in/cciep3
> >
> > _______________________________________________________________________
> > 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
> >
>
>
-- Shiran Guez MCSE CCNP NCE1 http://cciep3.blogspot.com http://www.linkedin.com/in/cciep3
This archive was generated by hypermail 2.1.4 : Fri Feb 01 2008 - 10:37:58 ARST