When using BGP there is no requirement to have equal bandwidth for load balancing.
BGP metrics, and BGP route determination do not have any thing that checks on bandwidth.
BGP RIB and IGP RIB are completely different Henry. IGP's such as RIP, EIGRP, OSPF, all have specific metrics and algorithms you can modify to tune it to your liking or leave it at the default. EIGRP has Bandwidth and delay by default, and OSPF has cost. BGP, left to it's default, makes RIB decisions based on shortest AS-Path. If you have multiple entries with the same AS-Path you can start manipulating various other parameters such as Weight (Cisco only), Local Preference, MED values, etc etc.
Are you load-balancing via eBGP with the same ISP or different ISPs?
MED can be used only if the ISP's support it hey!
I would still stick with local pref only if Henry has two ISP's and wants to load balance outbound traffic..
Simple answer is NO.
If you want to know more about load balancing/sharing in BGP then please see below links.
was just thinking about this one a while back... we could use IPSLA to force the BGP path to be different if one of the links is over loaded.. IPSLA parameters could be that the latency or even rtt..
No, they can be on entirely different bandwidth. May be one used for primary and other as a backup
Sent from Cisco Technical Support iPhone App