Link to home
Start Free TrialLog in
Avatar of cfan73
cfan73

asked on

DC connectivity – dynamic failover options (BGP, routing over GRE, etc.)

Customer is looking to implement dynamic routing to support auto-failover between DC interconnections. The main location has a P2P optical connection to the primary (local) DC where most of their compute/storage lives, and an MPLS path through a secondary DC in another city. The DCs are connected via a 10-Gbps backbone. Currently all routing is handled statically, which means that when there’s an issue with the primary connection, it’s a manual process to build routes and force traffic over the MPLS connection through DC2. They need to automate this failover, so it’d sound as simply as configuring dynamic routing everywhere.
The complications to do so would include:

- The MPLS network can only support dynamic routing through BGP peering (no redistribution of anything else between the CE and carrier PE devices).
- The DC infrastructure also only supports BGP peering, and it requires a dual-peer/connection handoff for SLA/redundancy reasons. (It would also cost the customer more in terms of services support and cross-connects in their DC/hosting contract.)

It would seem as if we could bypass the BGP requirements by configuring a GRE tunnel between the DC1 network and Corporate HQ, and route over this tunnel using standard EIGRP or OSPF.

My questions have to do with whether the GRE tunnel solution would create any limitations or concerns vs. fully configuring BGP on both ends. Again, the goal is to simply enable full failover to the MPLS circuit if/when the P2P optical circuit had an outage.

Thank you, and please let me know if additional information would be required/helpful.

User generated image
Avatar of Jan Bacher
Jan Bacher
Flag of United States of America image

if the core switch does not do BGP and connects to both locations, you should be able to do OSPF (or ISIS) between the three devices.

set a cost so that the physical PtP is preferred.
Avatar of cfan73
cfan73

ASKER

Thanks for your input. The core switch is Enterprise-licensed and would handle BGP (as would the cloud resources in the DCs on the other end), so that's not a concern. Plus, BGP is the only allowed routing hand-off into the MPLS PEs at either end, so we can't use OSFP, EIGRP or IS-IS.

So, what I'm really trying to get to is whether we'd have to move forward with the extra expense and design of deploying BGP, or if we could adapt a solution (such as a GRE tunnel) that would carry data transparently across the MPLS and avoid BGP altogether.
ASKER CERTIFIED SOLUTION
Avatar of Jan Bacher
Jan Bacher
Flag of United States of America image

Link to home
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
Avatar of cfan73

ASKER

Thank you for the input. I'm going to leave this open for additional thoughts/comments.