Would the latency between dallas to dallas or houston to houston datacenters be prohibitive to creating a "virtual private rack" that spanned datacenters?

Many of the features of a virtual private rack are very neat.

However, one thing the planet's size has allowed me is to have one server in one datacenter and another server in another datacenter for geographical, building, network, and IP redundancy.

Could virtual private racks span datacenters in the future via some planet fiber ring, etc. or would the geographicial latency make this less than ideal compared to having a cluster of servers all in one datacenter like with the present private rack offerings?

Likely yes, the latency involved between cities would cause a problem with a unified virtual rack environment... especially if you were routing through a Dallas firewall to a Houston server (for example).

We're running through a number of scenarios now and the most probable solution includes multiple VPR environments linked together over site to site VPN services (on the back end) and global load balancing (on the front). This will allow you a level of redundancy and interoperability while allowing us to offer the best possible functional solution.