Need BGP Experience Here!

[cscheper1]

I am wondering if clearing the bgp route tables will cause the cores to reevaluate the route options. My new 500M circuit is at 1% utilization while my 100M circuit in the same ASR peaks at 100% and I lose packets.

Logic tells me to let the protocol sort things out, but I'm impatient and want to try to fix this.  The scenario:

-We have 4 ISPs that are EBGP Neighbors that connect through 3 ASR1000 edge routers.

-We have an HA pair Nexus7000 core routers that are IBGP neighbors with the ASR1000.

-We have recently removed one ISP neighbor that had 100M and replaced it with another ISP neighbor with  500M.  Currently our AS neighbors are

ASR1000-1:  AS#XXX - 300M 

ASR1000-2:  AS#YYY - 200M

ASR1000-3:  AS#ZZZ - 100M

                      AS#AAA - 500M

The 500M connection to the neighbor AS#AAA is the new connection.  Only a few of the routes from AS#AAA are making it to the routing tables of my core Nexus.  I have checked different subnets manually and typically all of the neighbors have an as path that is the same number of hops (typically 3).  I am not keen on weighting or setting local preferences and I am wondering if clearing the bgp route tables will cause the cores to reevaluate the route options.  As it is, my new 500M circuit is at 1% utilization while my 100M circuit in the same ASR peaks at 100% and I lose packets.

Will clearing the bgp routes do anything?  Or will they rebuild the way they are now?

[Julio E. Moisa]

Hi

You should consider a BGP attribute like local preference to select the best path to reach the destinations, if you are receiving the same traffic through the 4 ISP. Also aspath attribute could be used to have a symmetric traffic.

It is just an example: 

500M - Local preference value 1000
300M - Local preference value 800
200M - Local preference value 600
100M - Local preference value 400

Clearing the bgp routing table should not be the solution. Also implement changes during maintenance windows to avoid any downtime. 

Hope it is useful

:-)

[cscheper1]

Julio, thank you!  Paul called you out for kudos in his response and expanded on my situation.  I wish I could tag you both with correct answer.

/shep.

[Julio E. Moisa]

Hi

No worries, Im happy to know you got the correct solution.

Have a good day

:-)

[paul driver · Accepted solution]

Hello

I am not keen on weighting or setting local preferences and I am wondering if clearing the bgp route tables will cause the cores to reevaluate the route options

To manipulate your egress/ingress traffic and load share between your asr1000's rtrs you will need to adopt using some bgp path attributes such Local-preference as Julio suggested for egress traffic and possibly as-path pre-pending for ingress traffic from your isp's

Leaving it to the mercy of the lower orders of the bgp path selection can provide you with such issues you are now facing.

res
Paul