Failover Between 2 Datacenters

[ken price <kprice75@xxxxxxxxx>]

All this chat between Joel, Joseph, and Roberto has let to some questions.  Is anyone aware of any project that would allow me the same level of failover/redundancy that an LVS project like Keepalived provides, but between clusters colocated with 2 different providers?

I'll go into more detail.  I have a full-on production HA Cluster "A" colocated with Exodus Communications in San Francisco.  I have a lower-powered backup HA cluster "B" colocated with Worldcom in Chicago.  Assuming I'm talking about static content (I'm not, but this is hardly the forum for database or filesystem replication over VPN), how can I failover to backup cluster "B" if San Franciso suddenly falls into the ocean.  I'd like to failover at LVS/Heartbeat speeds of 30-60 seconds max.  I'd rather failover at VRRP speeds of less than 5 seconds. 

Here's the kicker.  Because Worldcom can't route Exodus IP's, how would/could you handle this at the DNS level?  Could the answer lie in Dynamic DNS-type technology?  For the sake of this discussion, lets eliminate the possibility of both clusters living with the same Colo provider ... what would happen if Exodus or Worldcom suddenly went out of business?  :-)

-Ken

---

Do you Yahoo!?
The New Yahoo! Search - Faster. Easier. Bingo.