describes the time when a service is available for usage. If planned outages (maintenance windows) are counted as Downtime or not when calculating the achived availability depends on the SLA.

Availability is a hot topic when discussing the hardware and required budget. Two - or even three servers, active-active or active-standby cluster ? Hardware- and Software vendors provide impressive presentations about their products to provide "High Availability".

Experts do know that "High Availability" is not "Continous Availability", but project sponsors and other business users learn that difference after they question why IT requests permission for a planned outage / maintenance window although they paid so much for High Availaibilty.

As the additional costs for "continous availability" will in most cases not be justified the only way to boost the "Total Availability" by

reducing the number of outages to a minimum
reducing the duration of planned outages

To achive 1) you need to require that all (or most) configuration changes don't require a restart.
To achive 2) you need to require performance indicators for shutdown and startup and include them into your acceptance tests.

But this are only two of 150 non-functional requirements provided in our Template for Non-Functional Requirements !

Some are easy to achive, and you might (nearly) for free - you just should not foreget to ask for them. Some might cost more, but when writing the RFP you are still in a project phase where the budget is not frozen.

Asking later for adding one of those requirements will for sure cost much more.

You really should read those 150 non-functional requirements BEFORE you write your next RFP.