From my experience:
1.AppDynamics has some out of the box calibrations that it uses based on statistical formulae. THe health of a tier will depend on what you have associated with that tier, and what type of agent you have deployed. You can also customize health rules to meet the needs of your definitions (always a best practice in my opinion as each organization and applicaitons have different use-cases).
2. The status of a node is again dependent on the type of agent you have deployed (machine or server agent), and then the performance of said node. AppDynamics will compare performance against baseline data. There are several metrics that roll into this analysis, such as cpu, disk, and memory performance.
3. Dynamic baselines can use a few different variations. How this is applied depends on what areas you are specifically reviewing. But, you will see that the variations are typically varied by thte timeframe of data you are using in the configuration. The out of the box condition evaluation will use the time window that you have set.
4.When changing the time window, you are changing the focus of the analysis. Any change in the constraint of the time window will affect the calculation of the health against the baseline.
One thing to note, AppDynamics uses a statisical model that includes outliers more often than a typical standard deviation model. This means that we care more about the spikes and dips than the average. (at least from a data retention perspective).
I highly suggest you go through the basic training course. I put my staff and my internal "customers" through it, and they received a solid understanding of most of what you are looking for.
Also, keep in mind, I am not an employee... :)
... View more