topic Re: Failover between two DC's in Network Architecture & Design

Failover between two DC's

MLD — Fri, 19 Jul 2019 22:46:29 GMT

Hi Folks,

I currently have 2 DC's that are in the same geographic region along with several remote sites. All locations are connected via private Metro-E service with L3 connectivity (OSPF).

I am considering running iBGP w/OSPF redistribution to speed up failover between our DC sites, but would like any suggestions or comments regarding a DC-DR failover. The end goal being complete failover between DC-DR

Re: Failover between two DC's

Erik_Auerswald — Thu, 12 Sep 2019 15:11:27 GMT

Hi Matthew,

have you tried to use BFD to speed up OSPF convergence, i.e., time to detect a failed neighbor (Metro-E indicates that you have a Layer 2 service and currently use OSPF directly between your DC devices)? If you have several E-Lines instead of one E-LAN you could consider point-to-point link-type configuration to possibly speed up OSPF convergence.

Perhaps you could benefit from using OSPF areas, if you are concerned with two many routes affected by a failure in the other DC, or just to keep events localized in one DC if possible.

I am not yet convinced that the added complexity of using two routing protocols with redistribution is needed, but then I do not know much about your situation and what you have tried already.

Why do you think that using iBGP would speed up convergence? You seem to have an idea what it shall accomplish.

Thanks,
Erik

Re: Failover between two DC's

MLD — Thu, 12 Sep 2019 18:41:58 GMT

@Erik Auerswald

I was also considering multi-area OSPF to keep failures localized (as you stated). Our E-LAN is a mesh so the p2p configuration is not possible at this time.

I was considering iBGP as a future proof configuration, in the event that we start to expand to different regions and to speed up failure. However, the BFD with multi-area ospf might be a better option at the moment.

Thanks for your reply.