Backups bucket grows 200 GB/month. Nobody sorts. Lifecycle sends everything to Glacier after 30 days — including today's emergency restore backup. Estimated restore: 12 hours.
Lifecycle rules automate decay; written wrong, they automate panic.
Policy goals
Cut cost, legal retention, expire temps. Chain Standard → IA/Glacier → Deep Archive → delete.
Align with RPO/RTO per object class.
Prefix structure
Table by prefix: tmp delete fast, logs IA then expire, legal long hold, DB backups Glacier with noncurrent rules.
Versioning
Noncurrent transition/expiration. Abort incomplete multipart after 7d.
Mandatory tests
Shadow bucket, analyzer dry-run, quarterly Glacier restore drill. Billing anomaly alerts.
Multi-cloud
S3, GCS, Scaleway, OVH — syntax differs, pitfalls identical. IaC + PR review.
Quarterly lifecycle review
Export policies. Orphan prefix list. Cost per storage class trend. Glacier sample restore test. Legal hold tag check. Incomplete MPU abort count.
Adjust retention per legal/compliance audit.
Document owner per rule — avoid "mystery 2022 rule".
Governance
Owner per bucket policy quarterly review. Finance sign-off retention >1 year — Glacier retrieval cost included.
Tag data-class=regulated bypass lifecycle delete — automation respects tags.
Transition to Glacier before backup restore test: data loss policy not data — rule review board.
CloudWatch transition failure metrics — silent fail = Standard objects forever, bill surprise.
Lifecycle incident
Rule added delete 7d on wrong prefix — restore from versioning if enabled fast enough.
Emergency disable rule: documented lifecycle config suspend — not panic delete rule.
Cost anomaly detection: bill +30% month — check mistaken transition to expensive retrieval class.
Operational summary
Lifecycle automates cost and retention — test Glacier restore before rule, noncurrent versioning, legal hold tags. Quarterly prefix owner review.
Prefix mistake = data loss or bill surprise — shadow bucket tests.
Lifecycle workshop
Participants: ops, finance, legal, product. Output: prefix, retention, class, owner table. Shadow bucket test 2 weeks. Go-live rules with documented rollback disable.
Month 1 post-bill review: cost vs forecast.
Finance alignment
Lifecycle saves money month 3 not month 1 — transitions and retrieval fees learning curve. Finance expects instant drop — set expectations.
Legal hold exceptions must be tagged at upload, not discovered during litigation.
Cross-team workflow
New microservice onboarding checklist includes lifecycle prefix registration. Finance receives monthly storage class breakdown — catches transition misconfiguration early.
Legal hold tag overrides delete lifecycle rule — test tag application on sample object quarterly.
Transition to Glacier Deep Archive cheapest but retrieval hours — confirm RTO accepts.
Operational follow-up
Align lifecycle with HR document retention — avoid legal conflict. Finance validates Glacier retrieval cost before rule. Shadow bucket lifecycle test automates object class count compare — drift alert. Document gaps between host marketing and field measurement in the quarterly review.
Quarterly follow-up
Align lifecycle with HR document retention — avoid legal conflict. Finance validates Glacier retrieval cost before rule. Shadow bucket lifecycle test automates object class count compare — drift alert. Document gaps between host marketing and field measurement in the quarterly review.
Overlapping lifecycle rules need documented vendor evaluation order — conflicts cause surprises.
Keep a dated runbook, before/after metrics, post-incident review — cumulative discipline beats Friday night panic.
Keep a dated runbook, before/after metrics, post-incident review — cumulative discipline beats Friday night panic.
Keep a dated runbook, before/after metrics, post-incident review — cumulative discipline beats Friday night panic.
Keep a dated runbook, before/after metrics, post-incident review — cumulative discipline beats Friday night panic.
Keep a dated runbook, before/after metrics, post-incident review — cumulative discipline beats Friday night panic.
Keep a dated runbook, before/after metrics, post-incident review — cumulative discipline beats Friday night panic.
Deep dive object-storage-lifecycle
Keep before/after metrics for each change, the last successful drill date, and documented gaps between host marketing and field measurement. A quarterly review with business teams avoids repeating the same mistakes. To compare infrastructure and read more field notes, browse our directory, comparison tool, and technical blog guides.
Keep before/after metrics for each change, the last successful drill date, and documented gaps between host marketing and field measurement. A quarterly review with business teams avoids repeating the same mistakes. To compare infrastructure and read more field notes, browse our directory, comparison tool, and technical blog guides.:::note Key takeaway. Lifecycle = automatic contract on objects; test restore and noncurrent versions before prod. :::
Decide and move forward without blind spots
- Prefix / retention / class / owner table — ops, finance, legal, product workshop.
- Two-week shadow bucket — test transitions without touching prod.
- Handle noncurrent versions — separate versioning rules, abort incomplete MPU.
- Quarterly Glacier restore drill — time before writing prod rule.
- Month 1 cost review — finance forecast aligned (savings visible month 3).
Object storage: compare Scaleway, OVH, AWS via our directory and comparison tool.
Frequently asked questions
Lifecycle vs home cron?
Lifecycle = native, reliable, low ops. Home cron = flexible but mass-delete bug risk.
Instant Glacier transition?
No — minutes to hours. Glacier restore can take hours and cost; plan retrieval class.
Bad prefix?
Rule on logs/ may archive still-served objects if prefix too broad. Test on shadow bucket.
Versioning + lifecycle?
Handle noncurrent versions separately — common bill explosion.
Write the lifecycle rule after timing a Glacier restore — not before.