From: ccie2be (ccie2be@nyc.rr.com)
Date: Thu Dec 23 2004 - 18:41:07 GMT-3
Kurt,
Actually, when you're using a p2p - any p2p, not just a f/r p2p sub, there's
no need to map the remote L3 address to a local L2 address. This is because
with a p2p - any p2p - there's only one entrance and one exit.
Consider this. When you configure ppp on a serial interface, do you map the
remote ip address to any L2 address?
Anticipating your next question, why then is the frame interface-dlci <dlci
#> command needed?
The answer is so that the router can encapsulate the layer 3 packet with the
correct L2 headers and trailers. A dlci is frame-relay what a mac address
is to ethernet. By using the frame interface-dlci command, your telling the
router what L2 address to use when the router creates the L2 frame.
Also, don't forget, with f/r a physical interface may have multiple dlci's
assigned to it. Once the frame leaves the router and arrives at the frame
switch, if the dlci weren't there, the frame switch wouldn't know what pvc
this frame belongs to and therefore wouldn't know where to switch the frame.
In summary, here's what happens:
When a packet arrives, the router using the route table (or some faster
method) and the destination ip address, determines which interface the
packet needs to use to get to it's next hop.
If the interface is a frame relay p2p sub-interface, the router now has to
determine how to encapsulate the packet with the L2 headers & trailers. For
this, the router uses the info you provided with the frame interface-dlci
command.
If the interface is a frame relay p2m interface ( phy or sub, it doesn't
matter), then some L3 to L2 address resolution is needed. That L3 to L2
resolution can be either dynamic or static - either frame inverse-arp or
frame map.
HTH, Tim
----- Original Message -----
From: "Kurt VanDerMark" <kvandermark@highpoint.com>
To: "ccie2be" <ccie2be@nyc.rr.com>; <ccielab@groupstudy.com>
Sent: Thursday, December 23, 2004 4:07 PM
Subject: RE: Frame Relay Questions
Ok....so if you have a point-to-point subinterface, you specify the
dlci with the 'frame-relay interface-dlci xxx' command. You cannot use
frame-relay map statements with a point-to-point subinterface.
By using the frame-relay interface-dlci command, aren't you relying on
inverse-arp,
and hence dynamic mapping, to give you the ip address of the other end of
the link?
If you show your frame-relay map, won't this result in a dynamic entry?
If a requirement is to not use dynamic mapping, wouln't you have to change
that interface
type to multipoint so that you can disable inverse-arp and put in a specific
map statement?
I hope I'm not splitting hairs here...
-----Original Message-----
From: ccie2be [mailto:ccie2be@nyc.rr.com]
Sent: Thursday, December 23, 2004 10:51 AM
To: Kurt VanDerMark; ccielab@groupstudy.com
Subject: Re: Frame Relay Questions
Kurt,
I'll tackle your question about inarp.
This is how I understand it.
By default, inarp is enabled and so is LMI.
LMI is a managment protocol that runs between the router and the switch -
it's local to that link.
There are 3 "versions" of LMI and you don't have to run the same version at
each end of the pvc.
What LMI does is announce to the router what pvc's (dlci's, really) are
active on the interface.
Once the router knows what dlci's are active, the router can inarp on each
dlci. This means the
router is asking for the remote router's Layer 3
address. (Note, I said Layer 3 address, not ip address because the same
process can work for clns or ipx I believe).
When you disable inarp, you disable the either the asking for or responding
to portion of the inarp process - I always forget.
But, the result is the same. If inarp is disabled throughout, you
effectively disable dynamic mapping.
HTH, Tim
----- Original Message -----
From: "Kurt VanDerMark" <kvandermark@highpoint.com>
To: <ccielab@groupstudy.com>
Sent: Thursday, December 23, 2004 8:39 AM
Subject: Frame Relay Questions
> I'm doing a little frame relay review, trying to make sure I nail down the
basics...and I have a few questions.
>
> Please correct anything I say that's wrong...
>
> I understand frame relay can be configured on an interface 3 ways:
> - physical interface (considered multipoint)
> - subinterface point-to-point
> - subinterface multipoint
>
> Say for instance I have a connection going from one router defined as a
physical interface, to another router defined via a subinterface. To avoid
routing issues such as interface mismatch with IS-IS (any other issues or is
this the only one?), I'll want to define the subinterface as multipoint.
If the other end of the connection is defined by a physical interface, why
wouldn't I always define the subinterface end as multipoint? If you aren't
worried about routing issues, is there some advantage to defining that end
as point-to-point?? Are there some split-horizon issues?
>
> Also, as a best practice, I understand its best to disable inarp on an
interface if you plan to use map statements before bringing the interface
up. If you don't plan to use map statements, and want to use dynamic
maps, does the FR switch just notify the router about all the mappings it
knows about, or does the service provider do filtering on the switch, or how
does that work?
>
> Sorry if these are basic questions, but I want to make sure I have this
down...
>
> Thanks.
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
This archive was generated by hypermail 2.1.4 : Mon Jan 03 2005 - 10:31:29 GMT-3