Load Balancing between multiple BGP peer connections

Wed Sep 14 18:41:11 UTC 2005

On Sep 14, 2005, at 2:27 PM, Tom Sands wrote:

> Is the connectivity with 1 provider or 3 separate provider?  The  
> diagram and wording would seem like its the same.

Good point.  My post assumed all three links were to the same upstream.

-- 
TTFN,
patrick

> Patrick W. Gilmore wrote:
>
>
>>
>> [Wow, operational content - thank you!]
>>
>> On Sep 14, 2005, at 6:24 AM, Joe Shen wrote:
>>
>>
>>>
>>> How could load on multiple BGP peer links be balanced
>>> automatically?
>>>
>>> The situation we are facing:
>>>
>>>              ---------------------------|
>>>              |  Service provider        |
>>>              |                          |
>>>              --R1--------------------R2---
>>>                |\                     |
>>>                | \       E-BGP        |
>>>                |  \ ----------------\ |
>>>                |                     \|
>>>             ----R3--------------------R4-|
>>>             |     Our Network            |
>>>             |      (OSPF)                |
>>>             -----------------------------
>>>
>>> The three links between our network and Service
>>> provider network are all 1Gbps. Now, we noticed that
>>> load on link R1-R3, R1-R4 is about 50% (in/out), but
>>> load on R2-R4 are about 90% ( in/out).
>>>
>>> How could we balance those load on the three links
>>> automatically? or must we tune the route mannually?
>>>
>>
>>
>> It's not really clear whether the traffic is going mostly in or   
>> mostly out.  In fact, it looks like you are 1:1 balanced from the   
>> paragraphs above.
>>
>> Inbound traffic and outbound traffic need to be balanced  
>> separately.   I'll do inbound here, since that's slightly harder.   
>> If you need help  with outbound too, let us know.
>>
>>
>> First, what are you announcing to your SP?  Is it just one big   
>> block?  Or several?
>>
>> Second, does your SP accept MEDs?  (And if not, maybe you need a  
>> new  one? :-)
>>
>> If you have just one big block, split it up into multiple little   
>> blocks.  Announce them with MEDs or prepend on some links and not   
>> others or just announce a subset on each link.  Be sure to set   
>> NO_EXPORT on the little blocks, and announce the aggregate prefix  
>> on  all links.  You do not want the little blocks making it into  
>> the  global table (wouldn't do any good anyway).
>>
>> It's not difficult to do, it's just a bit of trial-and-error to  
>> get  right unless you have Flow Stats or something so you know  
>> where the  traffic is going anyway.
>>
>>
>> Outbound is a variation on this theme.  Let us know if you need   
>> outbound help as well.
>>
>> Good luck.
>>
>>
>
> -- 
> ------------------------------------------------------
> Tom Sands
> Chief Network Engineer
> Rackspace Managed Hosting                   (210)447-4065
> ------------------------------------------------------
>
>