But, I think some other points might need to be monitor,
for example, cpu usage, disk io, etc..; system event log, cluster log, etc..;
DB file growth, table space growth, etc..;
duration growthof jobs, index usage, health of data buffer, etc..
In addition, we can cycle errlog when finding errlog size is too big,
we can use logparser to filter the errorlog that is old history date before two days, to increase the ERRORLOG reading performance.