In this chapter
We'll meet lifecycle management — real, automatic rules that transition aging objects to cheaper storage classes, expire data past its real retention need, and clean up old versions and abandoned multipart uploads — closing the real cost questions versioning and multipart uploads both raised, with a deliberate, ongoing policy instead of manual cleanup.
The Problem in Real Life
Three months after enabling versioning, GreenMart's object storage bill has genuinely grown — old versions of product images nobody will ever request again, still sitting there, still being paid for in full. "We wanted the safety net," Mike says, "not a bill that grows forever."
Sarah nods. "That's a real, solvable problem — but it needs a deliberate policy, not someone remembering to clean up manually."
Old versions and old receipts are just sitting there, costing us real money. Can that be automatic?
Mike
Manual Cleanup, Eventually vs. A Real, Automatic Policy
Storage classes trade retrieval speed for cost
Cold/archive tiers cost far less than standard storage, in exchange for slower real retrieval — a fit for rarely-read aging data.
Real, automatic rules — no manual policing
Transition and expiration rules apply continuously to matching objects, closing the cost questions versioning and multipart uploads raised.
Lifecycle Management
Lifecycle management is a real, deliberate set of rules an object storage bucket applies automatically over time — moving objects to cheaper storage as they age, or deleting them outright once they're no longer genuinely needed — without anyone at GreenMart having to remember to do it manually.
- Storage classes — the same data, at a genuinely different real price. Most object storage systems offer multiple real storage classes for the exact same object model: a "standard" tier, priced for frequent access, and one or more real, cheaper "cold" or "archive" tiers, priced far lower in exchange for slower real retrieval time (sometimes minutes or hours, not milliseconds) — a real, honest trade of retrieval speed for storage cost, appropriate for data that's rarely, if ever, actually read again.
- Transition rules — moving data automatically as it ages. A real lifecycle rule can say, precisely: "move any receipt older than 90 days from standard to a cold storage class," applied automatically to every matching object, on an ongoing basis, with zero manual intervention. GreenMart's own receipts are a genuinely strong fit — read constantly in their first few weeks (customer support, returns), then almost never again, except for the rare compliance audit.
- Expiration rules — deleting data automatically once it's genuinely done. A real lifecycle rule can also say "permanently delete objects older than 7 years" — directly solving real, legitimate data-retention requirements, without anyone having to remember, or manually hunt down, exactly which objects have finally aged out.
- Closing two real, open threads from earlier chapters. A lifecycle rule can specifically target old versions (the previous chapter's own subject) — "keep only the 3 most recent versions of any object" or "expire non-current versions after 30 days" — directly answering the real cost problem versioning raised. And a lifecycle rule can automatically abort and clean up incomplete multipart uploads (two chapters back) left abandoned past a set period, closing that chapter's own honest, open cost concern too.
| Rule | Applies To | Real Effect |
|---|---|---|
| Transition after 90 days | Customer receipts | Standard → cold storage class (cheaper, slower to retrieve) |
| Expire after 7 years | Customer receipts | Permanently deleted, matching real retention requirements |
| Keep only 3 most recent versions | Product images (versioned) | Older versions automatically expired — bounds the real storage cost |
| Abort incomplete uploads after 7 days | All buckets | Abandoned multipart upload parts cleaned up automatically |
GreenMart now has the real, automatic policy layer this whole Act has been building toward: not a one-time migration, but an ongoing, deliberate, self-maintaining system that keeps storage cost proportional to what GreenMart genuinely still needs, without anyone manually policing it.
Key Takeaway
Lifecycle management applies real, automatic rules over time — transitioning aging data to cheaper storage classes, expiring data past its real retention need, and cleaning up old versions and abandoned uploads — turning storage cost management from a manual chore into a deliberate, ongoing policy.
Why This Matters
As GreenMart's own object storage grows — years of receipts, countless product image versions — lifecycle management is the real, deliberate mechanism keeping cost proportional to genuine, ongoing need, instead of every object GreenMart has ever written silently accumulating cost forever.
GreenMart now has real, automatic control over both versioning's cost and its own data's natural aging — transitions, expirations, and cleanup, all deliberate policy, not manual maintenance. The next chapter turns to a related but distinct real concern: avoiding storing the exact same content twice at all, through content-addressable storage.