Routing
Routing

bgp peer load sharing

‎06-04-2019 01:09 AM

Hello,

 

I'm trying to set up some type of load sharing for my bgp peering sessions that I have at exchange AMS-ix. Currently, I have 2 core routers (router1 & router2) with the same bgp peering sessions but at the moment all traffic from the sessions are going over router1. I would like to configure the bgp peering session in such a way that 50% of the session is active in router1 and other 50% over router2. I have read up on using MED and Multipath but this will not really achieve what I need, does anyone have a suggest or provide an article that I can read up would be much appreciated. Both paths between my core router's are of equal bandwidth

3 REPLIES 3
Routing

Re: bgp peer load sharing

‎06-04-2019 01:54 AM

Hi,

 

Hope this helps: 

https://www.juniper.net/documentation/en_US/junos/topics/example/routing-selecting-multiple-equal-co...

 

 

Regards,

Rahul Gautam

Please mark my solution as accepted if it helped.

 

Routing

Re: bgp peer load sharing

‎06-04-2019 03:13 AM

Also note that you can only control the traffic you send from your two peers to the IX.  The IX will decide based on their own configuration and policies how to use the dual peers to your network for inbound traffic.  It sounds like they are treating your pair of peers as a primary and secondary peering on their side.

 

Steve Puluka BSEET - Juniper Ambassador
IP Architect - DQE Communications Pittsburgh, PA (Metro Ethernet & ISP)
http://puluka.com/home
Highlighted
Routing

Re: bgp peer load sharing

‎06-04-2019 09:10 AM

Hello,

If you have ECMP multipath shoud help you to achieve BGP traffic load balancing:

https://www.juniper.net/documentation/en_US/junos/topics/topic-map/bgp-multipath.html

You can also use link-bandwidth community to perfrom unequal BGP traffic load balancing

https://www.juniper.net/documentation/en_US/junos/topics/example/bgp-multipath-unequal.html

Please note you will need multipath knob and per packet load balancing policy in both the scenoarios. 

Thanks & Regards
Vishal Singh