Adam Donnison hosted a session about how the *.mysql.com websites are architectured. He explaind how the situation was before they started building the current implementation and showed several drawbacks with that approach (Apache and MySQL on the same machine, no failover) and how they fixed it. He even showed how well the new MySQL Monitoring solution works without noticing that while he is showing live data, that the monitor reports a peak of requests and several red (dangerous) areas in the system. The site went effectively down when he explained how easy all this is to manage. He got very busy when we notified him, but he actually got everything up and running again within a few minutes.