Best Practice for Multiple Data Center Router Redundancy?

I understand VRRP can be used within a data center for redundancy but is less desirable across data centers. What are the best practices for router redundancy across data centers?