Storage Tiering as a FinOps Discipline

Storage costs creep silently. Tiering on access pattern is a 60-80% reduction with no app changes.

Why storage creeps

Hot data lives on hot tier; cold data accumulates on hot tier; never moves.

Without lifecycle, cost grows linearly with data; with lifecycle, cost grows logarithmically.

Four-tier model

• Hot (S3 Standard, GCS Standard).
• Warm (S3 IA, GCS Nearline).
• Cold (S3 Glacier, GCS Coldline).
• Archive (S3 Deep Archive, GCS Archive).

Access-pattern analysis

Pull access logs; classify each object by last-accessed; place on tier.

Most buckets show 60-80% of data unaccessed for >90 days.

Lifecycle policies

Automate via lifecycle policies; auto-tier based on age.

Manual tiering always falls behind; automated tiering compounds savings.

Antipatterns

• Manual tiering. Falls behind.
• Aggressive cold-tier without retrieval awareness. Costs of getting back exceed savings.
• One bucket without lifecycle. Cost compounds.

What to do this week

Three moves. (1) Apply this lever to your highest-spend workload. (2) Measure the dollar impact for one month. (3) Roll the practice out to the next two services if the savings hold.