10decodersAWS Cost Optimization, Done Without Downtime
A behavioral health services provider runs its core operations platform — client management, treatment workflows, scheduling, reporting, and compliance — on AWS. Its infrastructure had grown organically, and nobody had gone back to confirm which resources were still needed. 10Decoders ran a full cost assessment, then optimized the environment in careful, validated stages.
Cost creeps in quietly — this is how it was found and removed
The client's core operations platform runs on AWS infrastructure that had grown organically over time. Databases, compute instances, storage volumes, and virtual desktops were all running on flexible, pay-as-you-go pricing, and several resources were no longer in active use — but nobody had gone back to confirm which ones were actually safe to touch.
10Decoders ran a full cost assessment across the environment, then optimized it in careful, validated stages: reserved pricing where usage was already stable, and snapshot-and-batch removal where resources were unused — checking the client's application and data platform after every change before moving to the next one.
The AWS environment in scope
Production Database
Moved from on-demand to reserved pricing without a configuration change or downtime.
UAT & Production Compute
Assessed for reserved-pricing eligibility against real, observed usage patterns.
Storage Volumes
Unattached volumes identified, snapshotted, and removed in controlled batches.
Virtual Desktops
Unused WorkSpaces confirmed and decommissioned after client approval.
Data Platform
Job execution and table freshness monitored after every infrastructure change.
Ongoing Reporting
Utilization and reservation reporting set up for continued cost visibility.
Unused resources don't disappear on day one — they disappear once we're sureit's safe.
The Challenge
- Production database and compute instances ran on flexible, on-demand pricing despite stable, predictable workloads.
- Ten unattached storage volumes were quietly generating recurring charges with no one confirming whether they were safe to remove.
- Unused virtual desktops kept accruing storage costs long after anyone was using them.
- The same environment powered a live data lakehouse — any change had to be validated against real application and data-pipeline impact, not just billing.
Our Approach
- Assessed real usage against pricing models and recommended reserved pricing only where usage was already stable.
- Reviewed every unattached volume, took a snapshot first, then removed them in small, monitored batches.
- Confirmed which WorkSpaces were truly unused, secured client sign-off, then decommissioned them with evidence shared back.
- Monitored data-pipeline jobs and table freshness after every change before approving the next one.
The Solution
10Decoders built a phased, evidence-based optimization process rather than a one-time cleanup: assess, validate, implement in small batches, monitor, then move to the next opportunity only once the last one was confirmed safe.
Every destructive step — deleting a volume, removing a WorkSpace — went through client approval first and a snapshot second, with monitoring evidence shared back before the next batch began.
Assessment to safe, validated action
AWS Billing & Usage Data
RDS, EC2, EBS, and WorkSpaces utilization pulled directly from AWS billing and reservation reporting.
Assessment & Validation
Cost drivers identified, ownership and usage validated, and every change snapshotted before it's made.
Controlled Implementation
Approved changes rolled out in small batches, monitored, and confirmed safe before the next step.
Six habits that kept cost cutting from becoming risk
Evidence Before Action
Every optimization was sized against real usage data, not assumptions.
Snapshot-First Safety Net
Nothing was deleted without a recovery point already in place.
Small-Batch Rollout
Changes were made in controlled batches, not all at once.
Data-Platform Awareness
Job execution and table freshness were checked after every change.
Client-Approved Every Step
Nothing destructive happened without sign-off first.
Built for Ongoing Visibility
AWS billing and reservation reporting was set up to keep monitoring cost after the engagement.
Five workstreams, one coordinated cleanup
Infrastructure Cost Assessment
Full review of RDS, EC2, EBS, and WorkSpaces pricing models and utilization to find real, specific savings opportunities.
Reserved Instance Optimization
Recommended shifting the production database to 1-year reserved pricing, sized against its existing configuration.
EC2 Cost Optimization
Reserved-pricing recommendations for UAT and production compute instances based on stable usage.
EBS Storage Cleanup
Snapshot-first, batch-by-batch removal of unattached storage volumes.
WorkSpaces Decommissioning
Identification, client approval, and removal of unused virtual desktops, with evidence shared back.
What changed
RDS pricing reassessed
Production database identified for a ~17.6% cost reduction through reserved pricing, with no configuration change.
EC2 pricing reassessed
Reserved-instance savings identified across UAT and production compute.
Unattached storage cleaned up
Ten unused EBS volumes validated and progressively removed in controlled batches.
Unused virtual desktops removed
Two unused WorkSpaces decommissioned after client approval.
Infrastructure risk contained
Every destructive change was preceded by a snapshot and followed by monitoring.
Ongoing visibility established
AWS billing and reservation reporting set up for continued cost tracking after the engagement.
How the engagement was run
Delivery & Governance
- Phased assess → validate → implement → monitor methodology.
- Client approval required before every destructive change.
- Snapshot-first policy on all storage changes.
Transparency & Evidence
- Continuous status updates and supporting screenshots.
- Databricks job and table-freshness validation after every change.
- Documented before/after comparison for every optimization area.
Reusability & IP
- A repeatable cost-optimization model applicable to additional AWS resources.
- AWS Reservation Utilization reporting set up as an ongoing monitoring mechanism.
- Framework designed for future infrastructure reviews, not a one-time fix.
Impact across the organization
Lower Recurring Cloud Spend
Clear, validated savings across compute, database, and storage without introducing new operational risk.
Continuity You Can Trust
Every change was validated against the live application and data platform before the next one proceeded.
Savings You Can Verify
A clear line between identified and confirmed savings, backed by documented evidence at every step.
