HSRP and EIGRP across two data centers

[Wan_Whisperer]

I am modifying an existing network work design to allow full IP redundancy across two networks using HSRP.

Ideally, I would not trunk VLANs across two data centers I would implement VXLAN to limit the broadcast domain. 

Take a look at my diagram.  I have this working at a larger scale in my lab (way more vlans) and everything works as desired.  

My question is what are the ramifications of advertising both networks via IGP (EIGRP) in both data centers?  I know more overhead but let's say that the primary router in Denver is the active router for VLAN 1 and a server on VLAN 2 located in New York has packets destined for a server on VLAN 1 located in Denver?  The packets hits its DF in New York and the primary router in New York is advertising VLAN 1 but the the path is better than Denver's path to VLAN 1, remember the active router is the primary router for VLAN 1. 

Thanks,

WW

[M02@rt37]

Hello @Wan_Whisperer,

while advertising both networks via IGP in both data centers can work, it may result in suboptimal routing and increased overhead. To ensure efficient traffic flow and prevent suboptimal routing, it's recommended to configure PBR or route maps to prefer the path through the active router for the destination VLAN.

[MHM Cisco World]

this hard task, two enterprise network interconnect, 
first VXLAN, VXLAN use for intra not inter DC

for your second Q, please can you more elaborate ?

[Joseph W. Doherty]

I believe some of the latest L2 technologies, such as @MHM Cisco World shows in his posting, deal with topology, much as L3 can.  Unsure if your VXLAN does, or can.

For "classical" L2, having the same virtual GW across two WAN sites, you do have the common issue of two hosts, at the same location, on two such L2 networks, might have their transit traffic bounce off a remote site, rather than staying local.

If this is a problem, and you're using HSRP, on some HSRP implementation, you can have multiple virtual GWs, hosted on the same device, same network, yet, if needed, they can backstop each other.

e.g. HSRP GWs

site1  192.168.1.1 primary, 192.168.1.2 secondary

site2  192.168.1.1 secondary, 192.168.1.2 primary

Hosts at each site would use its site local HSRP GW address.

Works great if host GW are statically defined, but if hosts use DHCP, you have the difficulty of hosts getting their local GW IP.  (If DHCP servers are at both sites, assigned IPs, the "faster" offer, at each site, should be the IP used, and "faster" might be accomplished by DHCP server being local.