Exercise 1: Choosing a Pricing Model for a Fault-Tolerant Batch Job — Possible Solution
====================================================================

The workload described -- short-lived, fault-tolerant, safely
restartable if interrupted -- matches Spot pricing's own real
trade-off almost exactly. Spot instances offer real discounts of up
to roughly 90% off the On-Demand rate, in exchange for the real
possibility that AWS can reclaim the instance with only a short
warning if it needs that capacity back elsewhere.

On-Demand would work but leaves real, significant savings unused for
a workload that doesn't actually need guaranteed, uninterrupted
uptime. Reserved instances and Savings Plans are built for steady,
predictable, LONG-RUNNING baseline load -- committing to one for a
short-lived, bursty batch job would lock in a real ongoing cost for
capacity the job might only need briefly and irregularly.

ANSWER: Spot is the best real fit here. The job's own described
fault tolerance and safe-restart capability directly neutralizes
Spot's one real downside (the lack of an uptime guarantee), while
still capturing its real, substantial discount over every other
pricing model.

WHY THIS WORKS AS AN ANSWER
------------------------------
This correctly matches Spot's real risk profile (interruptible, no
SLA) against a workload description specifically engineered to make
that risk a non-issue, rather than recommending Spot generically --
and explicitly rules out Reserved/Savings Plans by noting they solve
a genuinely different problem (steady long-term load) than the one
described.