Zhivko Todorov
ALL CASE STUDIES

CASE 16 · LATTICE · 2025

S3INTELLIGENT TIERINGLIFECYCLEGLACIER

Four petabytes of S3, half the bill.

A geospatial analytics company had four petabytes of raster imagery in S3, growing 50TB a month. Everything was Standard. The bill was $94k a month and rising. We mapped actual access patterns, ran a lifecycle migration, and dropped storage cost 51% without touching application code.

INDUSTRY

Geospatial analytics

DOMAIN

COST

DELIVERED

2025

STACK

S3 STORAGE LENS·S3 INTELLIGENT TIERING·S3 GLACIER INSTANT RETRIEVAL·S3 GLACIER DEEP ARCHIVE·S3 LIFECYCLE·S3 INVENTORY

RESULTS

What changed, by the numbers.

STORAGE BILL

−51%

$94K → $46K / MONTH

DATA REORGANISED

4 PB

NO RE-UPLOAD

p50 ACCESS LATENCY

< 1s

NO REGRESSION

RETRIEVAL COST

+3%

WELL UNDER MODELED CEILING

HOW IT WENT

S3 Storage Lens gave us the truth: 78% of the imagery hadn’t been read in 90 days. 92% hadn’t been read in 180 days. The data was kept because customers might query it eventually — and a few did, occasionally, on the long tail.

We modelled lifecycle policies against the actual access histogram. Standard → Intelligent Tiering for the active 22%. Glacier Instant Retrieval for the 90-day tail. Glacier Deep Archive for anything not touched in a year. Crucially, no application code changed — the Lambda that fetched imagery already handled retries on RestoreObject.

The bill dropped 51% by month two. The increase in retrieval fees turned out to be 3%, well under our modelled ceiling of 8%. Customer-facing latency was unchanged. We left the team a Storage Lens dashboard and a quarterly review process so the same drift doesn’t happen again.

READY WHEN YOU ARE

Let's get your AWS bill (and architecture) in order.

The discovery call is free. You walk away with at least one concrete idea — even if we never work together.

Or email directly →