|
There are several options for the goal you are
seeking. The question is how much money you have. The best solution of course
would be to distribute the site over geographically distant locations,
multi-homed at each site peering (BGP) with multiple providers. Also, don't
forget farms of redundant servers/switches/routers at each site with load
balancers.
I suggest starting your high redundancy plan with a
single good data center with redundant connections to an ISP with a very high
SLA and a redundant backbone that is peered to multiple providers. Go redundant
on switches and routers to your server farm, and add a pair of load balancers.
You can also add DNS load balancers, cisco calls theirs "distributed director".
I personally don't believe in relying on DNS for disaster recovery, unless I
have a data center explode, I can bring everything back up long before the DNS
updates propagate everywhere.
----- Original Message -----
|
