If a service consists of multiple components prone to failure, the service’s overall availability depends on whether the components are in sequence or in parallel.
In sequence
Overall availability decreases when two components with availability < 100% are in sequence:
Availability (Total) = Availability (Foo) * Availability (Bar)
If both Foo and Bar each had 99.9% availability, their total availability in sequence would be 99.8%.
In parallel
Overall availability increases when two components with availability < 100% are in parallel:
Availability (Total) = 1 - (1 - Availability (Foo)) * (1 - Availability (Bar))
If both Foo and Bar each had 99.9% availability, their total availability in parallel would be 99.9999%.