Zeptrix CloudPrune guide
AWS cost reduction without making production risk an afterthought
A useful cost review does more than identify a large number. It shows the resource, supporting evidence, expected saving, owner, blast radius, safer implementation path, and how the result will be validated after a change.
1. Establish the billing baseline
Use Cost Explorer or the Cost and Usage Report to separate recurring baseline spend from one-time charges and recent anomalies. Compare services, accounts, regions, usage types, and tags before investigating individual resources.
2. Collect read-only evidence
Inventory resources and combine billing data with CloudWatch utilization, attachment state, lifecycle configuration, ownership tags, architecture context, and recent deployment history. A low CPU graph alone is not a deletion decision.
3. Rank the work
Prioritize recommendations using monthly impact, evidence confidence, implementation effort, reversibility, downtime risk, and owner availability. High theoretical savings with unclear production impact should not automatically move first.
Compute and commitments
Start with waste you can prove
Review stopped or consistently underused EC2 instances, unattached EBS volumes, old snapshots, idle load balancers, and unassociated public IPv4 addresses. For active workloads, compare rightsizing and scheduling before considering architectural migration.
Savings Plans and Reserved Instances can reduce stable baseline costs, but purchasing a commitment before removing waste can lock in the wrong capacity. Measure hourly usage stability, account sharing, coverage, utilization, and business growth assumptions first.
Data, storage, and observability
Retention and sizing decisions need operational context
RDS and Aurora reviews should include CPU, memory pressure where available, connections, storage behavior, replica purpose, Multi-AZ requirements, backup policy, maintenance windows, and application failover behavior. A quiet database may still be critical.
For S3 and CloudWatch, separate ingestion, storage, retrieval, query, metric, and retention charges. Shorter retention may help, but reducing noisy log production or changing a high-volume metric can produce a larger result. Preserve compliance and incident-response requirements.
Networking
Trace the path behind transfer charges
NAT Gateway cost combines hourly gateway charges with processed data. Review route tables, availability zones, traffic destinations, S3 and DynamoDB gateway endpoint opportunities, supported interface endpoints, and cross-AZ paths. A cheaper route is only useful if DNS, endpoint policy, security group, and availability behavior remain correct.
Calculate realizable savings
State the current cost, proposed cost, implementation cost, commitment or migration assumptions, and confidence range. Do not add overlapping recommendations twice—for example, scheduling and rightsizing the same instance.
Prepare rollback before change
Record the previous configuration, snapshot or export requirements, traffic-draining steps, maintenance window, validation metrics, stop conditions, and accountable owner before touching production.
Prove the result afterward
Verify service health immediately, then compare normalized usage and cost over an appropriate billing window. Record realized savings separately from the original estimate and feed the result into the next review.
How CloudPrune fits
Turn AWS signals into an accountable operating plan
AWS Cost Explorer, Compute Optimizer, Trusted Advisor, and Cost Optimization Hub provide valuable native signals. CloudPrune brings those signals together with inventory, telemetry, ownership, risk, rollback, and workflow context so a team can decide what to do next.
CloudPrune begins with read-only discovery. Recommendations can then move through evidence review, dry-run planning, explicit approval, controlled action, and post-change verification.
AWS Well-Architected Cost Optimization guidance provides the underlying AWS design principles used throughout this review approach.