MPLS PE-CE with EIGRP

Answered Question
Jan 10th, 2010

Dear Friends,

I am testing MPLS PE-CE with EIGRP as the Routing protocol. Here I observed a different behaviour of EIGRP. Let me brief the scenario

CE1-----PE1------P------PE2-------CE2

I am exporting CE2 route 200.200.200.200 into PE1 using MBGP. Its Metric is :2297856. Till this point to confusion. The same Metric value :2297856 I am seeing in PE2 & PE1 both VRF and BGP VPN table.

Then I am redistributing this BGP prefix 200.200.200.200 inside the EIGRP vrf address-family process so that CE1 will learn about this prefix. I am seeing the Metric value now changed to 2809856. I changed the metric value seed parameters in the redistribution command also to see the change.

  address-family ipv4 vrf SAIRAM-1
  redistribute bgp 100 metric 1 1 1 1 1
  network 10.0.0.0
  autonomous-system 1
  exit-address-family 

But the result is same and Metric value is not changing from 2809856. So Two things i want to get clarifed

I am seeing the value 0x8800:32768:0 in the output of show  "ip bgp vpnv4 vrf SAIRAM-1 200.200.200.200". Thus is clear that the route is INTERNAL & same AS as PE1

1. Is this the normal Behaviour in MPLS with EIGRP or IOS change is requrired???

2. Why I am not able to change the Metric of the said prefix going to CE1 even after changing the metric value in redistribute cmd.

3. Since the Route is Internal & EIGRP AS of CE1 & CE2 is matching why it is not using the extened community

Thanks for reading this patiently and I hope you will help me to understand this logic

sairam

I have this problem too.
0 votes
Correct Answer by Giuseppe Larosa about 6 years 10 months ago

Hello Sairam,

some notes about the pdf attachment from your tests.

a) EIGRP router for 200.200.200.200/32 is of type internal on CE1

CE1#show ip route 200.200.200.200
Routing entry for 200.200.200.200/32
Known via "eigrp 1", distance 90, metric 2809856, type internal

b) it is correct that metric changes from the point of view of CE1: EIGRP adds the delay on the path and checks for the smaller BW on path.

as we see on the following lines of  CE1#show ip route 200.200.200.200

Route metric is 2809856, traffic share count is 1
Total delay is 45000 microseconds, minimum bandwidth is 1544 Kbit
Reliability 255/255, minimum MTU 1500 bytes
Loading 1/255, Hops 2
CE1#

if we look at the parameters that are carried inside MP BGP and received by PE1:

Cost:pre-bestpath:128:2297856 (default-2145185791) 0x8800:32768:0
0x8801:1:640000 0x8802:65281:1657856 0x8803:65281:1500

if we calculate the difference of the two metric values:

2809856 - 2297856 = 512000 = 256 * 2000

this 2000 should account for the delay on serial interface from the point of view of CE1 in sh int ser0/0 we should see 20000 microseconds and delay is measured in tens of microseconds.

This is correct the metric from the point of view of CE1 cannot be the same as seen on PE1.

c) in a working solution you should redistribute EIGRP into BGP address-family ipv vrf SAIRAM-1 also to allow the local site to be seen on the remote site.

Hope to help

Giuseppe

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 5 (2 ratings)
Loading.
Giuseppe Larosa Sun, 01/10/2010 - 05:23

Hello Sairam,

because this is a lab I would suggest to post the configurations of PE1 and PE2

also post show  ip bgp vpnv4 vrf SAIRAM-1 200.200.200.200

what IOS images are running on PE1 and PE2?

if As-number under address-family under router eigrp is set to 1 to both sides on PE1 and PE2 you should see the route as internal I agree on this.

verify that you are sending-community extended under address-family  vpvn4.

About setting the metric in the act of redistributing into AF in EIGRP I never tried this and I'm not sure it is possible, because all the efforts of EIGRP as PE-CE are meant to preserve original metric components as sent by remote PE using extended communities with MP BGP.

The EIGRP data structure is rebuilt using the information carried within the BGP extended communities

>>  I am seeing the Metric value now changed to 2809856

do you see this on the CE or you are using sh ip eigrp topology on the PE node?

Hope to help

Giuseppe

snarayanaraju Sun, 01/10/2010 - 07:24

Hi Giuseppe,

Thanks for helping. For your kind reference I am attaching the running config. Important details are highlighted to make your work easy.

sairam

snarayanaraju Sun, 01/10/2010 - 07:31

Hi Guiseppe,

Thanks for helping.

For your kind reference I am attaching the Running Configuration of PE1 & CE1 (single file contains both the config). To make your work easy, I have highlighted the important details in different colour

regards,

sairam

Attachment: 
Correct Answer
Giuseppe Larosa Sun, 01/10/2010 - 14:43

Hello Sairam,

some notes about the pdf attachment from your tests.

a) EIGRP router for 200.200.200.200/32 is of type internal on CE1

CE1#show ip route 200.200.200.200
Routing entry for 200.200.200.200/32
Known via "eigrp 1", distance 90, metric 2809856, type internal

b) it is correct that metric changes from the point of view of CE1: EIGRP adds the delay on the path and checks for the smaller BW on path.

as we see on the following lines of  CE1#show ip route 200.200.200.200

Route metric is 2809856, traffic share count is 1
Total delay is 45000 microseconds, minimum bandwidth is 1544 Kbit
Reliability 255/255, minimum MTU 1500 bytes
Loading 1/255, Hops 2
CE1#

if we look at the parameters that are carried inside MP BGP and received by PE1:

Cost:pre-bestpath:128:2297856 (default-2145185791) 0x8800:32768:0
0x8801:1:640000 0x8802:65281:1657856 0x8803:65281:1500

if we calculate the difference of the two metric values:

2809856 - 2297856 = 512000 = 256 * 2000

this 2000 should account for the delay on serial interface from the point of view of CE1 in sh int ser0/0 we should see 20000 microseconds and delay is measured in tens of microseconds.

This is correct the metric from the point of view of CE1 cannot be the same as seen on PE1.

c) in a working solution you should redistribute EIGRP into BGP address-family ipv vrf SAIRAM-1 also to allow the local site to be seen on the remote site.

Hope to help

Giuseppe

snarayanaraju Sun, 01/10/2010 - 21:18

Ho Thanks for your efforts to explain the concept. In fact I also calculated this in the same fashion and kept that as my undertstanding and was waiting for approval. Now I got the confirmation from you experts and thus I am moving this to my study note.

Further to this discussion, I also learned when there is a backdoor link which also advertise the same prefix, to avoid suboptimal routing POI is used which makes the routing decision based on the metric and not using AD. Thus path thru MPLS Backbone is always preferred instead of EIGRP backbone.

Is there any configuration option available to disable this feature. Of course I know that by changing the MBGP metric this can be achieved.  Is there any possibilities available to disable this behavior so that, AD will be preferred over metric    in this scenario

This may be trivial details to put more efforts to research on this topic. Still this is just a thought process

sairam

Giuseppe Larosa Tue, 01/12/2010 - 11:04

Hello Sairam,

I've just  reviewed the other thread about EIGRP as PE-CE protocol.

What do you mean with :

>> Further to this discussion, I also learned when there is a backdoor link which also advertise the same prefix, to avoid suboptimal routing POI is used which makes the routing decision based on the metric and not using AD.

I've found this reference to the use of site of origin to deal with backdoor links

http://www.cisco.com/en/US/docs/ios/iproute_eigrp/configuration/guide/ire_mpls_vpn_ps6441_TSD_Products_Configuration_Guide_Chapter.html#wp1057047

Are you referring to site of origin SoO BGP extended community attribute?

Hope to help

Giuseppe

Actions

This Discussion