Reliability is a property or level that system is generally functioning properly.
As an example an LDAP Server Implementation may answer to a "ping" but not be able to make a Success for a Bind Request.
Reliability implies there is a some level of Fault tolerance
Reliability for Data At Rest is often referred to a Durability
Instances continuously start, restart and stop because of failures, deployments or autoscaling. These operations makes an instance temporarily or permanently unavailable. For Load Balancing
should mark the instance as unavailable and Load Balancing
should skip unhealthy instances from the routing
as they cannot serve requests
There might be more information for this subject on one of the following: