Giuseppe,
it also depends on what failure you get.
If the network is down, traffic can't even hit the primaru ACE and therefore it can't redirect it to the backup DC.
One solution is often to use a GSS to answer DNS requests. The GSS can probe the ACE and see if it is alive before responding to the arp request with the correct ACE vserver ip address.
Another solution is the Route Health Injection.
Both sides inject a route in the network through ospf or whatever protocol you use and when an ACE goes down, the network redirect the traffic to the backup site.
Finally, you can also use a redirect on the ACE when all local servers are down but not the network.
For http, this should be a simply http 302 redirect.
For other protocols, you will need to create a rserver representing the backup ace and use it as a backup rserver with client nat.
Not very nice, because traffic will have to go through both sites.
Gilles.