The MPLS-OPS Archive

[Rajiv Asati <rajiva@cisco.com>]

Devendra,

Please see inline...

> suppose i
> forget these derived applications and talk about a basic MPLS cloud what
> would be the nature of traffic flow across it - that is the
> question...
> granted that the routing packets would
> be IP forwarded.

Not really.
If the routing protocol packet is destined to a non-directly connected
router, then the packet will very well be MPLS forwarded (assuming the
sending router does have a label available to the destination IP
address).

If the routing protocol packet is destined to a directly connected
router, then yes, you are right, it will be IP forwarded and
handled.

> > packet
> x needs to be stamped (or not)?
>
> I believe the most common case is that the packet is received on an
> interface which is a member of a VRF (yes, this implies MPLS
> VPNs).

Yes, but in general, the IP packet will become MPLS packet provided there
is an outgoing label available to that destination IP address.
(Of course, whether it will get sent out as an outgoing interface is
dependent upon the interface to be MPLS enabled.)


> > what
> trigger converts the potential push into actual?
>
> The packet needs to be sent on a core-facing interface with MPLS
> enabled. (This doesn't always happen - the packet could also be 
> sent
> on another interface which is a member of the same VRF, and thus
> would
> not need a label.)

Sure enough.
Also, if there is no label available to the destination IP address of
that packet, then the packet will be sent out as an IP packet, be the
outgoing interface MPLS enabled or not.


> > suppose
> there is an MPLS cloud - running IGP,MPLS and LDP and there
> > are no TE-LSP'S and NO VPN's...
> >
> > and within the cloud i ftp from router x to router y --will
> that
> > traffic be label switched?
>
> No, why should it?

No, if the ingress router (x) doesn't have the outgoing label to the
destination IP address of that ftp packet.
Yes, if the router x does.


Cheers,
Rajiv

At 02:20 PM 1/23/2005, Devendra.Vyas@relianceinfo.com wrote:

> Lotus notes not listening to me...i have again
> changed the settings....................hope this time its plain
> text..........
>
>
> The way im approaching the question is
> -
>
> Originally MPLS was developed for
> getting the speed of L2 and the ease of L3.although this original
> intention is no longer the main driver for deploying MPLS.....but
> TE,VPN's,COS are all DERIVED applications.
>
> suppose i forget these derived
> applications and talk about a basic MPLS cloud what would be the nature
> of traffic flow across it - that is the question...
> granted that the routing packets would
> be IP forwarded.
>
>
>
>
> Please don't use HTML for email on this
> mailing list.
>
> > so then how does MPLS decide whether -
> >
> > packet x needs to be stamped (or not)?
>
> I believe the most common case is that the packet is received on an
> interface which is a member of a VRF (yes, this implies MPLS VPNs).
>
> > what trigger converts the potential push into actual?
>
> The packet needs to be sent on a core-facing interface with MPLS
> enabled. (This doesn't always happen - the packet could also be 
> sent
> on another interface which is a member of the same VRF, and thus
> would
> not need a label.)
>
> > suppose there is an MPLS cloud - running IGP,MPLS and LDP and
> there
> > are no TE-LSP'S and NO VPN's...
> >
> > and within the cloud i ftp from router x to router y --will
> that
> > traffic be label switched?
>
> No, why should it?
>
>
> ------- The MPLS-OPS Mailing List Subscribe/Unsubscribe:
> http://www.mplsrc.com/mplsops.shtml
> Archive: http://www.mplsrc.com/mpls-ops_archive.shtml