Great discussion topic. !00% may be achievable for some duration only. But sooner or later...
We are currently looking at alternatives for better uptime for a web based customer facing systems. In fact we have excellent uptime, generally not using a 4 hour once a week planned maintenance window. But we have had some nasty unplanned outages due to software, and while back a problem in a SAN on which a production cluster runs. So, following the old Boy Scout motto, "Be Prepared" seems to be the wisest plan.
What experiences do folks have with log shipping and replication (or other solutions) for minimizing those unplanned outages and not impacting availibility when maintenance is required?