Much really depends on how "availability" and "reliability" are defined.
If you define availability narrowly, it might only refer to the fact that the server is powered on and the OS is working. A broader definition might include the fact that all services offered by the server are working properly. The broader definition might be used interchangably with reliability.
If you are using only the narrow definition of availability, you might need to use reliability as a seperate measure. In this case, you might define reliability as whether an "available" server is functioning properly in its assigned role. To measure this, you would need to monitor all services/applications that users rely on from a specific server. You would then need to gather statistics on the percentage of time that services (as opposed to the server itself) are available.
An example might be an email server. The server itself and its OS might be rock solid (i.e. 100% availability - using the narrow definition), but the email software installation might be totally corrupted (which would equate to 0% reliability).
How is your team defining these two measures of performance?
