We come pretty close. We run multiple apache servers, which are being loadbalanced by a Foundry ServerIron cluster (2 loadbalancers with fallback). All customer data is on a NetApp storage cluster with about a petabyte of data, each head can take over the disks of an other head. The only thing that is not fully H/A is MySQL. We run it in replication mode, so the main SQL server is being replicated to 2 other servers, but it needs manual intervention. MySQL Cluster just doesnt feel right yet for us.