Frame Relay (Theory):
1. Talking about how having a physical interface to pass through a frame relay cloud (a switch). In real life, a router is used as frame relay switch (router can be configured into a switching mode.
2. Two modes of frame Relay
2a. Point to point (For a different subnet environment)
2b. Multipoint (For same subnet environment)
4. Frame relay makes your medium of connectivity transparent (you can't see it)
5. Virtual circuits creation out of a single medium (FR switch)
6. Frame relay (layer 2 protocol) does not contain IP addresses.
+ Apply DLCI number (because frame has no MAC address)
+ DLCI number are given by your ISP.
7. Frame routing = Frame switching.
8. Frame relay signaling (LMI)
Cisco proprietary =
ANSI = International standard
ITU = International standard
+ The purpose of LMI is for keepalive (To figure out whether a link is working or not)
9. You can apply a particular speed (commited VC speed) onto your virtual circuit (pros)
This is because it is logical (you can manipulate the bandwidth)
10. If you didn't manage the bandwidth properly, you might suffer a possible bottleneck. Cisco has this thing called "congestion management"
The sender has to find a way to figure out whether the path ahead is congested or not.
BECN = Backward explicit congestion notification
FECN = Forward explicit congestion notification
BR = Branch
FR = Frame Relay
Path 1 Path 2
HQ -------------------------------- FR -------------------------------------------BR
FR would know that if Path 1 is congested by analysing traffic coming to it.
If path 1 is congested.
BECN = 0
FECN = 1 (path 1 congested)
As traffic reaches BR with encapsulation of BECN = 1 and FECN = 0, BR will then send back traffics to HQ with the following encapsulation:
BECN = 1 (Path 1 congested)
FECN = 0
When HQ receives the traffics (with all encapsulation), it will know that Path 1 is congested and thus reduce traffic my 25%.
--> This will happen automatically.
11. In frame relay point-to-point (P2P)...
Logically all routers are connected in a serial mode (point to point)
Similar to "slip horizon" of RIP and IGRP, you can only have one way traffic , to overcome loops.
--> Cater for convergence or collision management. (Refer to the theory of stub network)
To counter split-horizon problem, you have to create virtual tunnel or link (to cater for bi-directional sort of scenario)
12. Point to point (mapping) is done automatically because eventually everyone will get it.
Point to multi-point mapping have to be done manually.
13. Frame Relay is not outdate, it is meant for the following benefits:
+ Cost benefits (without Frame Relay you probably need to have a few leased lines to connect to branches)
+ More security ( not making use of IP addresses (layer 3) even though still go through the bandwidth)
+ More efficient (faster)
14. PVC / PVI (In your streamyx)= Frame Relay identifier.
15. Configuration of Frame Relay will only be covered in CCNP.
16. In a frame relay network (point to multipoint), the IPs for all devices has to be in the same network. (same subnet)
17. In the real work, frame relay cloud would be located in the Central Office (CO)
Ethernet cannot use PPP because of the encapsulation problem (has to use PPPoE). Frame Relay is also another encapsulation (for serial interfaces as well), but it can work on ethernet.
Frame Relay Configuration (Practical):
if you are having multiple encapsulations within your network configured to all interfaces, then your serial link will be going up and down.
Thus, you need to make sure that all serial interfaces having the same encapsulation.
Point to Point Configuration: (Practical)
1. Apply frame-relay as encapsulation to serial interfaces.
2. Create sub-interfaces to serial interfaces.
3. Apply an IP address to the particular sub interfaces.
4. Apply DLCI number to sub interfaces.
5. Ping neighbors nodes to check if it is working.
6. Enable RIP(version 2) for network. Otherwise, the routing table is not there yet. (Now you have a fully routable frame-relay network)
7. Advertise for adjacent networks.
8. Also advertise for ethernet network. (have to wait for 40 seconds)
Create Multipoint (practical):
Not purely broadcast (probably just some portion of it)
1. create only one sub-interfaces and set it to multipoint.
2. Apply an IP address to the sub interfaces.
3. Do mapping. (Map local DLCI number to a particular IP address)
4. Ping your neighbors.
5. But you are unable to ping yourself (because your frame-relay don't have yourself in the entry, thus can't route back to you) --> why can't this be automatic.