04-30-2014 07:16 AM - edited 03-04-2019 10:53 PM
Hello,
I have a conundrum on my hands. I'll do my best to explain this simply.
We have 2 datacenters (I'll call them BP and CG). Between these datacenters we have a private circuit that connects them.
At each datacenter, we have a WAN router that goes out into AT&T's AVPN cloud. We have 20 or so remote sites that also have AVPN routers. All of their traffic goes through the AVPN cloud to the datacenters.
At each datacenter, we have a router advertising subnets via OSPF to the AT&T router. The AT&T router then redistributes these routes into the cloud via BGP.
The problem is that everything goes through the CG router. If traffic wants to get to BP, it goes through CG and over our private circuit. This adds a couple extra hops to the route.
The ideal situation would be this: if traffic needs to get to BP, then it would go directly to the BP router. If traffic needs to get to CG, then it would go directly to the CG router.
We would like this to be as dynamic as possible. If we add a new subnet, we don't want to manually enter a preferred or static route. Is there some kind of metric we can use to achieve this? Is our goal achievable using OSPF and BGP?
The attached diagram should help to visualize what I'm trying to communicate.
Any help would be appreciated!
04-30-2014 07:52 AM
You can tune OSPF and or BGP on each data center router to have a lower metric for routes that originate in that specific data center. This would eliminate transit traffic from going through your private circuit, unless one of the data centers looses it's link to AT&T.
04-30-2014 08:45 AM
Can you summarise each DCs internal sunbnets separately ?
If you can could the summary ranges also have spare capacity for new subnets ?
Jon
Discover and save your favorite ideas. Come back to expert answers, step-by-step guides, recent topics, and more.
New here? Get started with these tips. How to use Community New member guide