Exercise 3: Why a Scale Set Alone Doesn't Survive a Datacenter Outage — Possible Solution
====================================================================

A Virtual Machine Scale Set, on its own, guarantees that a defined
number of instances are running and load balanced, and that unhealthy
instances get replaced -- but nothing about that basic guarantee, by
itself, says anything about WHERE those instances are physically
located. Microsoft's own real documentation states this directly: "a
scale set alone can't protect you against data center failures."

If every instance in the scale set happens to be running in the same
single datacenter (or the scale set was never configured to spread
across zones at all), then a real failure of that one datacenter
takes out every instance simultaneously -- the scale set's own
replacement and load-balancing mechanisms don't help if there's no
surviving healthy instance anywhere else to fail over to.

Real Availability Zone distribution has to be EXPLICITLY configured
on top of the scale set -- it isn't an automatic, default behavior of
creating a scale set at all. And per Chapter 1's own real gotcha, this
protection is only available in the first place in regions that
actually support Availability Zones -- not every Azure region does.

ANSWER: This assumption is incorrect because a scale set's own basic
guarantees (replacing unhealthy instances, load balancing) say nothing
about physical distribution across datacenters by default. The team
needs to explicitly configure the scale set to distribute instances
across Availability Zones -- and first confirm, per Chapter 1, that
their specific target region actually supports Availability Zones at
all, since not every Azure region does.

WHY THIS WORKS AS AN ANSWER
------------------------------
This correctly separates what a scale set guarantees automatically
(instance count and health) from what it does NOT guarantee by
default (physical zone distribution), and explicitly ties the fix
back to Chapter 1's own real region-support gotcha rather than
treating "just enable Availability Zones" as always possible
everywhere.