HPE Community read-only access December 15, 2018
This is a maintenance upgrade. You will be able to read articles and posts, but not post or reply.
Hours:
Dec 15, 4:00 am to 10:00 am UTC
Dec 14, 10:00 pm CST to Dec 15, 4:00 am CST
Dec 14, 8:00 pm PST to Dec 15, 2:00 am PST
Comware Based
cancel
Showing results for 
Search instead for 
Did you mean: 

Help with BGP design

 
drolfe
Valued Contributor

Help with BGP design

Hi all,

 

I thought I would ask the qeustion as I'm very new to bgp design. Currently I have a IRF stack of hp 5900 switches, all client vlans and SVI's live on the IRF stack. I then have a default gateway from the DC, all static routing so very simple.

 

I"m wanting to start peering over an internet exchange with our 2 x /24 networks (Provider Agg).

 

The way I'd like to run things to it have the default gateway still route out to the DC HA core via the 5900's

 

Then any routes that are avalible via the internet exchange go via the peering router. I selected a single router for this as I always have the DC HA core to fall back on for default routing.

 

I assume at this stage that both the 5900 stack and BGP router would likely peer with both DC core routers. I really don't need the full table just yet as full transit will stay on the DC core for now.

 

What I don't really understand is how the 5900's will learn what routes are avalible via the bgp router as the internet exchange has around 150,000 prefix routes currently and the device can only handle 16K.

 

I was thinking to run the HP 5900 DC Baggs in a seperate VRF then under normal conditions send everything to the BGP router first or Run VRRP on the gateways between the Juniper and the 5900's ?

 

I've attached a diagram of how things would be patched, so just really looking for everyones thoughts on the best way forward.

 

 

Regards, Daniel

1 REPLY
becos
Senior Member

Re: Help with BGP design

Hello,

 

As far as I understood you are using the DC IP1.&2 as default gw and probably you have some loabalancing between them.

 

If your main issue is how the 5900 will hold the 150k Inet routes, well you can avoid using them a L3 hops and just have two new vlans cut from DC IP.1 to IP.4 and from DC IP.2 to IP.4 peering over iBGP if you can handle the extra L2 latency and b/w for L3 traffic from IRFs to DCs and then back down to eBGP peering router and also implement some kind of balancing for the eBGP  routes as well.

 

Another way is to redistribute eBGP into your existing IGP on IP.4 but that will depend on the kind of IGP and/or if you are using MPLS in your design.

 

Hope I gave you some hints.

./becos