FR addressing

[Guest_imported]

Hi,

i am a student preparing a FR overview for my colleagues and i have crawled the web to find dozens of white papers and for-dummies-tutorials just to end up with lots of questions, of which i want to address one right now, namly the addressing scheme of FR.

From what i've read, a DLCI has local significance only and is used in conjunction with a routing table filled with IP-Subnets. So, i assume that i tell the FR router to pass packets with the destination subnet address e.g. 9.0.0.0 on to the local DLCI, let's say, 50. Ok, data packets will be leaving my router on DLCI 50 and will arrive at the next FR switch. (BTW: I read that FR switches have routing tables, is this true? I thought only routers could have them..) Anyway, what happens at this router/switch then? My assumption is that the router unpacks the FR packet, looks at Layer 3 information (IP-Destination-address), looks up its own information of how to get to the IP-destination-subnet and will send the packet out on a port where this subnet can be reached at. Well, the problem is, if this is right, why is it called a "permanent" virtual channel and not a "I will find a way" channel. Many texts say that the path is known prior to sending the packet out and this is contradictory to finding a way on-the-fly. So may it be that there is a global address for an FR-router? One text stated that a DLCI has global significance since 1990 when Cisco et al. changed the existing FR standard, has this something to do with this?

Any cleaning up of my confused thoughts are very welcome. I guess the more i dig the internet the less i understand

Many greetings!

Bartman

 

[mbhandari]

Hi,
ONCE PACKET LEAVES THE ROUTER , FRAME-RELAY Switch it looks for DLCI Number . Frame relay switch switches the DLCI and forward the packet to destiend Interface.
-bhandari

 

[talisker]

Hi,

We're talking about two different stages here. First of all, the router considers a frame circuit as just another directly-connected interface with an IP address of 10.1.1.1

In the routing table is an entry that says: to get to network 10.1.2.0, the next hop is 10.1.1.1. So when a packet needs to get to network 10.1.2.0, the router sends it to 10.1.1.1 which is our serial interface (or sub-interface) with the frame circuit. The routing protocol knows that next hop is 10.1.1.2 (the other end of the frame circuit), so it sends it there, not knowing or caring that the packet will traverse multiple frame-switches to get there. The frame-circuit topology is invisible to the routers. At this point the packet is "encapsulated" into a frame-relay packet with a header that contains the DLCI number (ie DLCI 21). This DLCI is the ID number of the serial port on the router on the other end of the frame-circuit.

Next, the packet is sent to the first frame-switch which could have hundreds of PVCs (belonging to dozens of companies) defined on it. The DLCI in the packet header is read by the frame-switch and the switch-logic says "for DLCI 21, send (switch) the packet to interface serial3. The packet is then sent to the next frame-switch (connected to Serial 3)where the same process takes place. This continues until the frame-switch which is directly-connected to your company's router on the far end of the frame-circuit is reached.

Finally, the packet is sent from the last frame-switch over the T1 (or whatever your dedicated frame-relay circuit is)to your destination router's serial port configured as DLCI 21. Here, the frame-relay encapsulation is stripped off, and the routing protocol takes over. It reads the destination subnet (10.1.2.0) from the packet, looks in the routing table & sends it on it's way.

Please keep in mind that this is a simplified explantion. Many other things happen to the packet on the way - the circuit ID of DLCI 21 is only used between the (source) router & the first frame-switch. At this point a different identifier is used to send the packet to it's destination.

The PVC is considered "permanent" because the packets will always take the same "path" to get to the destination router. SVC is non-permanent because the "path is created each time traffic is sent, so packets may take a different path each time.

I hope this information is useful.