We help IT Professionals succeed at work.

Check out our new AWS podcast with Certified Expert, Phil Phillips! Listen to "How to Execute a Seamless AWS Migration" on EE or on your favorite podcast platform. Listen Now

x

BGP load balance between two ISPs

trojan81
trojan81 asked
on
Medium Priority
2,054 Views
Last Modified: 2012-05-11
Hello Experts,

I'm looking for suggestions. In my case study scenario, a company has two internet routers which have bgp peering to separate ISPs.

RTR-A peers to ISP-A
RTR-B peers to ISP-B
The two routers have an IBGP connection together.

The Routers are connected to a Firewall which does the PATing. For simplicity reasons, let's say there are only 2 internal network subnets. All routes from ISP-A has a local preference of 800 and all routes from ISP-B has a local preference of 700 so the internal network is taking ISP A.

If I want subnet 2 to take ISP-B, what are some suggestions to handle this?
Comment
Watch Question

Top Expert 2010

Commented:
http://www.cisco.com/en/US/tech/tk365/technologies_configuration_example09186a00800945bf.shtml

Take a look at the section: Load Sharing When Multihomed to Two ISPs Through Multiple Local Routers

Load balancing is not possible in a multihomed environment with two ISPs. BGP selects only the single best path to a destination among the BGP paths that are learned from different ASs, which makes load balancing impossible.

Since you want to route via source IP (Second Internet Subnet), you will have to implement Policy Based Routing.

http://www.cisco.com/en/US/products/ps6599/products_white_paper09186a00800a4409.shtml
http://www.cisco.com/en/US/docs/ios/12_2/qos/configuration/guide/qcfpbr_ps1835_TSD_Products_Configuration_Guide_Chapter.html

Billy
Hi,

Which firewall do you have?
SouljaSr.Net.Eng
CERTIFIED EXPERT
Top Expert 2011

Commented:
Load Balancing is in fact possible with BGP. Check this article out:

http://www.nil.com/ipcorner/LoadBalancingBGP/
I agree with rfc1180. Load balancing is not achievable. Load sharing is, to some extent. His link has got specific config example for load sharing. Assuming trojan81 has got no AS of his own, so he will have no control over incoming traffic. He can only distribute outgoing traffic.

I am not sure if trojan81 has got valid cisco contract, you might want to take a look at Cisco OER (Optimized Edge routing), which is bound to solve his issue.

Best,
Top Expert 2010

Commented:
>Load Balancing is in fact possible with BGP
Soulja, with due all respect, No, you cannot load balance utilizing BGP in his application; BGP selects only the single best path to a destination among the BGP paths that are learned from different ASs. The author's application, the BGP peers are in different ASes, if they were in the same AS, then yes, you could implement maximum-paths so that both routes are injected into the RIB. "If", the author wanted to "load balance"  the only option he would have is to set a better metric for the routes in the range 1.0.0.0 to 128.0.0.0 (More load sharing than load balancing) that are learned from ISP(A) and a better metric for the rest of the routes that are learned from ISP(B). However, he wants all traffic from one subnet (The second) to be routed via ISP-B, this is source routing and Policy Based Routing is required.

Author

Commented:
i appreciate everyone's input. I have been out sick the past couple days and will look into the suggestions tomorrow.
SouljaSr.Net.Eng
CERTIFIED EXPERT
Top Expert 2011

Commented:
Hi rfc,

I stand corrected.  :-)

Author

Commented:
RFC,

great link you gave me about load sharing.

http://www.cisco.com/en/US/tech/tk365/technologies_configuration_example09186a00800945bf.shtml


In this last example: Load Sharing When Multihomed to Two ISPs Through Multiple Local Routers

R6# show ip bgp
BGP table version is 15, local router ID is 192.168.64.6
Status codes: s suppressed, d damped, h history, * valid, > best, i - internal
Origin codes: i - IGP, e - EGP, ? - incomplete
   Network          Next Hop            Metric LocPrf Weight Path
*> 10.10.10.0/24    192.168.63.3                           0 300 100 100 i
!--- This line shows that network 10.10.10.0/24 is routed through AS 300
!--- with the ISP(A)-R1 link.

Why is it that R6 sees "300, 100, 100, i" as the best path to get to 10.10.10.0/24?  Prepending the AS towards ISP-A should have made the 10.10.10.0/24 network less desirable. So shouldnt R6 best way to get to 10.10.10.0/24 be "400, 100, I" through ISP-B?

Author

Commented:
RFC,

I re-read the example and see Both R1 and R2 are prepending AS 100 one time for network 10.10.10.0/24.
If both routers are prepending one time, what makes ISP-A the path to take from R6's perspective for network 10.10.10.0/24?
It sounds like the prepending cancels out
 
Top Expert 2010
Commented:
Unlock this solution and get a sample of our free trial.
(No credit card required)
UNLOCK SOLUTION

Author

Commented:
Thank you.
I am still undecided on which bgp path selection criteria to use for load sharing, but you have given me a better understanding.  
Top Expert 2010

Commented:
BGP path selection is not a criteria to use for load sharing; you are NOT able to do anyoutbound  BGP load sharing based on a source prefix from your network. You typically can influence on what transit circuits are to be used for inbound to your network by prepending, utilizing Metrics, or the more perferred way is to contact your ISP and ask the for their communities for changing the local_pref for the prefixes that you are advertising. Most ISPs will have this routing policy in affect and operational.

Billy
Unlock the solution to this question.
Thanks for using Experts Exchange.

Please provide your email to receive a sample view!

*This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

OR

Please enter a first name

Please enter a last name

8+ characters (letters, numbers, and a symbol)

By clicking, you agree to the Terms of Use and Privacy Policy.