cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
Announcements
924
Views
0
Helpful
8
Replies
Highlighted
Beginner

BGP Outbound Load Balancing

Hi everyone,

I'm having BGP configured on a pair of 6880-X in VSS, and we are peering with 3 peers and 2 AS. 

For a while now, I noticed that while our inbound are somewhat LB between 2 peers (on on each AS, i'm not sure why the 2nd peer on the same AS not picking up any traffic), outbound traffic only stick with one peer at a time.

I saw a lot of posts online, and in here, states to use LocalPreference to gain Load Balancing. But even when I apply LocalPreference to one peer, that I accomplished was outbound traffic now head out to another peer, but no LB at all.

 

What should I do? 

 

8 REPLIES 8
Cisco Employee

Hello Hieu, Can you please

Hello Hieu,

 

Can you please provide a simple diagram?

 

Regards,

Minh

Beginner

Here is a quick diagram (i

Here is a quick diagram (i have to change the AS #, sorry guys :) )

and also my BGP config

 

router bgp 4444
 bgp router-id 10.26.0.150
 bgp log-neighbor-changes
 network 64.79.131.0 mask 255.255.255.192
 network 64.79.151.0 mask 255.255.255.224
 network 66.116.101.0 mask 255.255.255.224
 network 66.116.119.96 mask 255.255.255.224
 neighbor 64.79.159.77 remote-as 23
 neighbor 64.79.159.77 soft-reconfiguration inbound
 neighbor 64.79.159.77 prefix-list BGP_DEFAULT_ONLY in
 neighbor 64.79.159.77 prefix-list BGP_LOCAL_ONLY out
 neighbor 64.79.159.89 remote-as 23
 neighbor 64.79.159.89 soft-reconfiguration inbound
 neighbor 64.79.159.89 prefix-list BGP_DEFAULT_ONLY in
 neighbor 64.79.159.89 prefix-list BGP_LOCAL_ONLY out
 neighbor 199.189.117.121 remote-as 22
 neighbor 199.189.117.121 soft-reconfiguration inbound
 neighbor 199.189.117.121 prefix-list BGP_DEFAULT_ONLY in
 neighbor 199.189.117.121 prefix-list BGP_LOCAL_ONLY out
 maximum-paths 5

 

Cisco Employee

Thanks for the info.So you

Thanks for the info.

So you want to share load for AS 4444. On the router in AS4444, please show ip bgp x.x.x.x where x.x.x.x is a prefix to be loaded sharing.

 

Regards,

Minh

Load Sharing When Multihomed

Load Sharing When Multihomed to Two ISPs Through a Single Local Router

http://www.cisco.com/c/en/us/support/docs/ip/border-gateway-protocol-bgp/13762-40.html#conf4

This is a very interesting thread. I'm building a similar network in a home lab. I located this article,which suggests "Load Sharing" as a possible solution. From what i can tell, the idea is to set a better metric for the routes in the range 1.0.0.0 to 128.0.0.0 that are learned from AS23 and a better metric for the rest of the routes that are learned from AS22, through the use of route-maps. 

One question, is the speed of all three links the same? 

Regards,

Patrick 

 

 

Beginner

Hi Minhda,it's  not any

Hi Minhda,

it's  not any particular prefix. I just want the outbound traffic to share out instead of just using one single best path.

 

Patrick,

all link has the same speed. That was an interesting concept, but it is still a little bit forcing and odd setup. 

I found a cisco article on this earlier, though I couldnt understand how they can get LB out from what they said. I'll put it up here in the next post.

after looking at my earlier

after looking at my earlier post I don't like the idea of denying prefixes, because if one circuit goes down then all of those prefixes are unavailable on the second link. 

here's another idea. instead of denying, each  range of prefixes is weighted higher per link, but it doesn't deny them from other links if link 1 goes down.

I hope this makes sense. I'm a student so i find this to be a cool topic. Maybe one of the More Experienced Cisco reps can look at the thread and lead us in a good direction. 

 

access-list 1 permit 0.0.0.0  63.255.255.255
access-list 2 permit 64.0.0.0 127.255.255.255
access-list 3 permit 128.0.0.0 255.255.255.255 


route-map A permit 10
match ip address 1
set weight 100
route-map A permit 20
match ip address 2
set weight 90
route-map A permit 30
match ip  address 3
set weight 80 
route-map A permit 40


route-map B permit 10
match ip address 1
set weight 90
route-map B permit 20
match ip address 2
set weight 100
route-map B permit 30
match ip  address 3
set weight 80
route-map B permit 40 


route-map C permit 10
match ip  address 1
set weight 80 
route-map C permit 20
match ip address 2
set weight 90
route-map C permit 30
match ip address 3
set weight 100
route-map C permit 40


router bgp 4444
no sync 
no auto-summary 
neighbor 64.79.159.77 route-map A in 
neighbor 64.79.159.89 route-map B in
neighbor 199.189.117.121 route-map C in

Hello Hieu,Have you made any

Hello Hieu,

Have you made any progress?  

Contributor

Hi Hieu, Can you post what

Hi Hieu,

 

Can you post what the BGP_DEFAULT_ONLY prefix-list is?

 

Also, can you post 'show ip route' and 'show ip bgp' for one (or all) of the prefixes permitted by the prefix-list?

 

In the meantime, keep in mind that the maximum-paths BGP feature for load balancing only works with the "same destination" which IIRC means the same AS.  

 

So, if AS 22's route is better than either of the AS23 routes, it would seem to me that (without any of the show ip output I requested) only the AS22 route would get installed. I'm assuming here that you are really just receiving a default route from the three peers for simplicity.

 

Lets see the show output and see if something doesn't jump out at us as the likely source of the problem.

 

-Tim

CreatePlease to create content
Content for Community-Ad
July's Community Spotlight Awards