A city is designing compute for a always-on citizen portal baseline…

Question

A city is designing compute for a always-on citizen portal baseline plus a fault-tolerant batch that transcodes public meeting videos. Which purchasing mix best aligns cost with workload behavior?

Accepted Answer

Cover the steady portal baseline with Savings Plans (or Reserved Instances where appropriate) and run the interruptible transcoding fleet on Spot Instances with a restart-friendly architecture. — Steady lights-on portal hours love Savings Plans; flaky-but-restartable video jobs love Spot. Locking RIs onto a weekend-only transcoder—or putting the whole citizen site on naked Spot—is the expensive (or embarrassing) mix-up.

Answer

Buy three-year All Upfront Reserved Instances for the entire transcoding fleet even though jobs only run a few nights per month.

Answer

Run both the portal and transcoding exclusively on On-Demand with no commitments and no Spot, ignoring predictable baseline hours.

Answer

Place the citizen portal only on Spot without capacity backups so every interruption takes the public site offline.

A city is designing compute for a always-on citizen portal baseline plus a fault-tolerant batch that transcodes public meeting videos. Which purchasing mix best aligns cost with workload behavior?

Short Explanation

Full Explanation