> I find that having a single point of failure with a backup > that can be manually brought in to place (such as an > asynchronously replicated database) is

> I find that having a single point of failure with a backup > that can be manually brought in to place (such as an > asynchronously replicated database) is quite often more > reliable than fancy home-made SAN solutions. As someone who has invested heavily in implementing HA systems and then seen them be the root of increased downtime, I have come to the same conclusion. A simple fail-over system that will result in short downtime & require limited manual intervention is in many cases the best route.