We had another outage (short) yesterday afternoon. This time the CDA (Content Display Application) cluster (jboss) had more connections than it could handle. It is the sort of problem you want to have (kinda). At least this time, the machines can handle more connections since the XSLT-caching was put into the system. Solution for now is to up the connections (conservatively to 100 each up from 75) and purchase another Apple XServe to add into the cluster.
We’re doing really well at handling the traffic that is normally coming to the sites. Now we just have to make changes so we can handle the traffic spikes. This is a bit of a moving target since as our sites grow the high water mark will too.