A healthy database should spend about 95% of its time waiting for CPU or IO, while an unhealthy database spends about 95% of its time in either waiting for network, SOS, MISC, and latch.
To summarize, if SQL Server is waiting on CPU and/or IO, then it is constrained by the hardware that is being used, and not by anything else. If the waits are caused by other things (networks, latches), then the queries being executed on the database aren’t “Healthy”. That is to say, it should be possible to tweak the queries to provide better performance.
