Is it just me...
or isn't it possible to use applications that are geographically redundant without resorting to sh...I mean stuff like this? Back in the day with Exchange we used to advise our clients against clustering through the OS because it wound up causing more problems than it solved. Eventually they moved their redundancy to the application layer and our clients are finally humming along with redudancy that finally works.
I recall one client recently who was sure OS clustering was still the way to go with some bastardized solution that sounds an awful lot like some of this stuff... and wound up with a flaky, almost unsupportable infrastructure.
Maybe I'm just old school, but the whole rush towards many tiny little OS'es with the redundancy/fail-over duct-taped underneath an OS and application layer that has no idea what's going on just seems like insanity.
Then again, I'm not too hip with server infrastructures outside of e-mail... so maybe it's just me.