EIGRP Point-to-Point Bandwidth Pacing

27 Comments

1. i am looking for the sol. of this goal.
   please provide the video solution

   1. Hopefully I can record them this week!

2. Sample config:

   Tilburg:
   on each subinterface bandwidth 25

   On spoke routers:
   bandwidth 64
   ip bandwidth-percent eigrp 1 19

   (25*0.5=12.5 12.5/64=19)

   Correct?

3. Hi Rene,
   Any idea when the video solution for this lab will be available ?

4. Hi Rene…

   Can you please post the configuration videos

   Thank you

5. I would set the s0/0 bandwidth on Tilburg to 25kb (256/10=25) and 25kb on every spoke router.

   Is this correct?

   Taken from Cisco website:

   Pure Multipoint Configuration (No Subinterfaces)
   In this configuration EIGRP will divide the configured bandwidth evenly across each virtual circuit. You must ensure that this will not overload each virtual circuit. For example, if you have a T1 access line with four 56K VCs, you should configure the bandwidth to be 224Kbps (4 * 56Kbps) in order to avoid dropping packets. If the total bandwidth of the virtual circuits equals or exceeds the access line speed, configure the bandwidth to equal the access line speed. Note that if the virtual circuits are of different capacities, the bandwidth must be set to take into account the lowest capacity virtual circuit.

   For instance, if a T1 access line has three 256Kbps VCs and one 56Kbps VC, the bandwidth should be set to 224Kbps (4 * 56Kbps). In such configurations, putting at least the slow virtual circuit onto a point-to-point subinterface is strongly recommended (so that the bandwidth can be raised on the others).

6. + add 58% percent bandwidth so each pvc will use around 32kb

   [quote=bartsmajdor]I would set the s0/0 bandwidth on Tilburg to 25kb (256/10=25) and 25kb on every spoke router.

   Is this correct?

   Taken from Cisco website:

   Pure Multipoint Configuration (No Subinterfaces)
   In this configuration EIGRP will divide the configured bandwidth evenly across each virtual circuit. You must ensure that this will not overload each virtual circuit. For example, if you have a T1 access line with four 56K VCs, you should configure the bandwidth to be 224Kbps (4 * 56Kbps) in order to avoid dropping packets. If the total bandwidth of the virtual circuits equals or exceeds the access line speed, configure the bandwidth to equal the access line speed. Note that if the virtual circuits are of different capacities, the bandwidth must be set to take into account the lowest capacity virtual circuit.

   For instance, if a T1 access line has three 256Kbps VCs and one 56Kbps VC, the bandwidth should be set to 224Kbps (4 * 56Kbps). In such configurations, putting at least the slow virtual circuit onto a point-to-point subinterface is strongly recommended (so that the bandwidth can be raised on the others).[/quote]

   1. This is correct. I haven’t had time to record this lab yet, I’ve been very busy…

7. HI Rene,
   like to see solution for this lab,

8. Hi folks,

   In Tilburg:
   Set the bandwidth to 256 kbps in serial 0/0.
   Totally 4 pvc’s so 256/4= 64 kbps per PVC.
   Set the bandwidth percent in Tilburg’s each serial SUB interface to 50%.
   50% of 64kbps = 32kbps(50% of CIR Achieved)

   Example configuration for only one subinterface(plz do it for all 4 subinterface according to the topology given)
   int serial 0/0.1 point-to-point
   ip bandwidth-percent eigrp 1 50

   In the Spoke:
   Set the bandwidth to 64kbps.
   Set the bandwidth percent to 50%.

   50percent of 64kbps = 32kbps

   That’s it.

   I Hope am right. Plz correct me if i am wrong. Cheers

   1. If we have 10 PVC 64 kb each

      Can we set 64/2*10 to the bandwith of hub router ?

9. Hi Rene,

   please provide the video solution

10. Was there ever a video solution for this lab created???

11. The interface bandwidth command will not affect traffic patterns at all. It’s just a reference for IGP protocols. In a real world situation you wouldn’t want to set an interface bandwidth command to change the IGP traffic calculations. To reduce congestion, you have to reduce traffic on the circuit. Therefore, you can remove broadcast, reduce hello packets, frame-relay congestion commands, and change EIGRP traffic percentages. I was able to test this theory by applying a service policy of 8kbs on the interfaces. If you ping the interface continously, you will see the queuing buckets fill up and drop packets. After reducing the congestion, you will notice a change in the traffic pattern. You will actually notice less packets being dropped. You will only notice this after adding the broadcast commands on frame-relay and EIGRP, and then changing the circuit to unicast only traffic + EIGRP traffic percentages.

    Below is an example configuration:

    policy-map ingress-policy-8kbps
    class class-default
    police cir 8000
    exceed-action drop

    int se0/0
    ip address 192.168.1.5 255.255.255.0
    encapsulation frame-relay
    frame-relay map ip 192.168.1.1 501
    service-policy input ingress-policy-8kbps

    router eigrp 1
    neighbor 192.168.1.1 se0/0
    no auto

12. i am looking for the sol. of this goal.
    please provide the video solution

13. Let me know if my solution is correct.

    on the hub router
    int s0/0
    bandwidth 256

    on the spokes
    (since there are 10 routers, the BW for each spoke’s serial interface is 256/10 = 25 (approx.). I chose 25 coz it will be the best BW allowed for each spoke router

    int s0/0
    bandwidth 25

    then for point-to-point BW pacing goal

    since we are required to use 50% of the CIR, hence 64(.5) is 32kbps..
    then use the ip bandwidth-percent on the pt-to-pt interfaces of the hub router and the s0/0s of each spoke router.

    for hub router

    int s0/0.x point-to-point
    ip bandwidth-percent eigrp 1 (percent)

    the percentage here is

    (what percentage?) of 25k is 32k..by simple math that gives you, 128%

    so that is
    note 10 pt-to-pt interfaces on hub router

    int s0/0.x point-to-point
    ip bandwidth-percent eigrp 1 128

    then for spoke router

    int s0/0
    ip bandwidth-percent eigrp 1 128

    hth,

    alvin

14. You are setting bandwith on a serial interface of the hub router, but this action is incorrect. If you are entering "sh int serial 0/0.101", you can see that bandwith of this subinterface not changed and equal 1544 kbit (the speed T1). The subinterface does not inherit bandwith the serial interface. EIGRP will be used bandwith the subinterface equal 1544. I think, you need set bandwith on every subinterface of hub router.

15. Thanks for your reply petrovich. I labbed this one out and you are right and I lacked configuring BW on the sub-if.

    So the configuration will be

    on hub router:

    int s0/0
    bandwidth 256
    int s0/0.1 point-to-point
    bandwidth 25
    ip bandwidth-percent eigrp 1 128
    :
    :
    int s0/0.10 point-to-point
    bandwidth 25
    ip bandwidth-percent eigrp 1 128

    while on each spoke routers:
    int s0/0.1 point-to-point
    bandwidth 25
    ip bandwidth-percent eigrp 1 128

    ~ alvin

    1. hey alvin,

       why not

       int s0/0.1 point-to-point
       bandwidth 32
       ip bandwidth-percent eigrp 1 100

       on spoke router??

16. My Solution :

    As per point to point interfaces you need to divide the bandwidth configured on the interface of the hub on how many PVC you have.

    if you will go by 4 PVC, you will have :

    256/4 = 64 kbps

    Set 64 kbps to the interface and subint of the hub.
    so far all interface has the same 64kpbs speed, with this
    eigrp will not overwhelm each pvc.

    now if you will go by 10PVC, you will have :

    256/10 = 25kbps

    set 25kbps to the interface and subint of the hub.
    but 50% of 25kbps is not equal to 50%of 64kbps pvc speed, which is one of the requirement of the problem.

    no worry, as eigrp bandwidth percentage can exceed 100%.
    we can set the hub to use at least 128% which is equal to 32kpbs which is 50% of PVC.

    problem solved. 🙂

17. I am confused – the topology shows 192.168.1.0/24 for all routers but in the lab configs Tilburg has the following
    interface Serial0/0.102 point-to-point
    ip address 192.168.12.1 255.255.255.0
    frame-relay interface-dlci 102
    !
    interface Serial0/0.103 point-to-point
    ip address 192.168.13.1 255.255.255.0
    frame-relay interface-dlci 103
    !
    interface Serial0/0.104 point-to-point
    ip address 192.168.14.1 255.255.255.0
    frame-relay interface-dlci 104
    !
    interface Serial0/0.105 point-to-point
    ip address 192.168.15.1 255.255.255.0
    frame-relay interface-dlci 105

    1. Tpeel,

       Do not go by the addressing in the topology. Make sure to check all subinterfaces on the Hub router for correct addressing and make sure neighbor relationships are FULL. The Hub should have 4 adjacencies (one to each spoke), while each spoke should have one (to the hub).

       1. I understood about sub-interfaces and the IP addresses set – I was pointing out the diagram showed 192.168.1.0/24 as the address range. I am working on OSPF over Frame-relay 😕 at present and getting my head around that 🙂

          Thanks for your help
          Tim 😀

18. I think it is 2 solutions.
    1. On hub with bandwidth 25kb/each subinterface. Rest for eigrp 12.5kb
    On spoke with bandwidth 64kb, for eigrp rest 32kb.
    On hub to bring at 32kb/link we need to raise width 32/12.5= 2.56, and 256%*50%=128%
    On spoke we let them untouched.
    Verify : eigrp bandwidth percent 128% * 25 kb = 32 kb.
    But 32kb * 10 = 320 kb, more than 256kb physical bandwidth of the hub. So this solution is bad.
    2. On hub with bandwidth 25kb/each subinterface. Rest for eigrp 12.5kb
    On spoke with bandwidth 64kb, for eigrp rest 32kb.
    We need to bring down 32kb to 12.5kb. So 12.5/32 = 0.39, and 39%*50%=19%.
    On hub we let bandwidth-percent untouched.
    Verify: eigrp bandwidth-percent 19 % * 64 = 12.6 exactly like hub. I think this is correct answer.

    1. Yours make sense.

       However in my opinion, i see it this way.

       10 spokes x 56K bandwidth each.
       Hub line 256K bandwidth.

       We don’t want each PVC on HUB to become over subscribed (for EIGRP traffic), so:
       256 / 10 = 25.6K

       HUB:
       int ser0/0.1 point-to-point
       bandwidth 25
       ip bandwidth-percent eigrp 1 50

       Keep going for the next 9 point-to-point interfaces on HUB as like shown above.

       SPOKE:
       int ser0/0.1 point-to-point
       bandwidth 25
       ip bandwidth-percent eigrp 1 50

       So in this scenerio, (25×50) / 100 = 12.5, eigrp won’t use more than 12.5K of bandwidth.

       Source:
       http://www.cisco.com/c/en/us/support/docs/ip/enhanced-interior-gateway-routing-protocol-eigrp/13672-12.html#hybrid

    2. Correction, each spoke has 64K instead of 56K per my previous note. Just replace 56K with 64K, and it should be OK. sorry for the confusion.

19. In my opinion it’s simple. 256/10 = 25.6*(0.5) = 12.8Kbps. So you have to insert the bandwidth command on hub interface s0/0 “bandwidth 13”.

    I did 12.8/64 = 0.2 or 20%. So the spokes routers should be configured with “ip bandwidth-percent eigrp X 20″and the “bandwidth 64”.

Comments are closed.