Exercise 2: Durability vs. Availability, With Concrete Examples — Possible Solution
====================================================================

Definitions, per the chapter:

DURABILITY means the data itself won't be lost -- it's still safely
stored somewhere, intact, even if you can't currently get to it.

AVAILABILITY means the data is currently reachable when requested --
you can actually retrieve it right now.

These are separate claims: data can be perfectly durable (still
existing, uncorrupted, safe) while being temporarily unavailable, and
in principle a system could even claim high availability for data that
turns out not to have been durably stored (though a well-designed
system aims to guarantee both).

Example incident affecting DURABILITY:

A storage service suffers a catastrophic hardware failure across
multiple replicas at once (an extremely rare event given the "11
nines" replication the chapter describes, but the relevant category of
incident) and some objects are permanently, unrecoverably lost. This
is a genuine durability failure -- the data itself is gone, not merely
temporarily unreachable.

Example incident affecting AVAILABILITY without touching durability:

A cloud provider has a temporary regional outage -- a networking issue
or a service disruption within one region -- that makes a storage
service unreachable for a period of time. Customers cannot retrieve
their files during the outage, but once the provider resolves the
issue, all the data is still there, completely intact, exactly as it
was before. The replication underneath the durability guarantee was
never actually compromised -- only the ability to REACH the data
during the outage window was affected.

WHY THIS WORKS AS AN ANSWER
------------------------------
The two examples are deliberately chosen to isolate each property: the
durability example shows data genuinely gone despite (hypothetically)
the service otherwise being reachable, while the availability example
shows data fully intact despite being temporarily unreachable --
matching the chapter's own tip-box point that "a temporary regional
service disruption affecting availability doesn't necessarily mean any
data was lost."