In my experience, when an eBGP session is established between customer and
provider, the session is built (all routes to be exchanged are initially
exchanged). Then incremental updates happen (routes are added/withdrawn) as
necessary. Most of the bandwidth utilization due to BGP will occur when the
peering is first established. This should not kill your bandwidth IMHO. In
a former lifetime, we once ran 6 full table eBGP connections (multihomed to
the same provider) on a Cisco 7507 router with 128MB of RAM and RSP4 route
processor card. This was several years ago when the full table was approx
50-60,000 routes per connection. Today it is about 90,000 routes.
"Gardner, Donald/COR" <[EMAIL PROTECTED]> wrote in message
C015C68E0018D4118B9A00508BC75E9A02070C25@rifle">news:C015C68E0018D4118B9A00508BC75E9A02070C25@rifle...
> OK, I have a discussion going on in the office about what a full BGP
session
> is going to do to our Internet bandwidth and memory.
>
> Right now I have 2 T1s from two providers, one from PSI and one from
Epoch,
> and am getting partial routes from both. I get 6505 prefixes from PSI and
> 1238 from Epoch. We are running a 7206 NPE150 with 128 Meg of Ram. I am
> looking at getting full routes from one or both providers.
>
> I am getting conflicting feedback about what getting full routes will do
to
> our bandwidth and memory. One says it will kill us and another says no
> problem.
>
> Any real world experiences and feedback would be appreciated...
>
> Don
>
> _________________________________
> FAQ, list archives, and subscription info:
http://www.groupstudy.com/list/cisco.html
> Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
>
_________________________________
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
