It writes all the nagios statistics like host state changes, service state changes, schedule downtimes, comments and so on in the database opmon.
All performance data is written to the datadase opperf and each host has a dynamic created table in the format: perf_ where host_id comes from the opcfg database. This strategy was tought to easy administration, increase query speed and update, instead of having a huge table, we split them in hosts.
In the opperf database there is a table service_metric that contains all the metrics a host and service have.
