2.6 Cost Optimization Strategy (New Solutions)
Official task mapping
SAP-C02 task 2.6: Storage tiering and pricing knowledge; right-sized cost and transfer modeling. This lesson applies the official task to an original design exercise; it is not an exhaustive exam syllabus.
Prerequisites
Review pricing models, S3 costs and network costs. Be able to explain the relevant failure boundary and access/cost model before choosing a service.
New-solution economics
Estimate compute, storage, requests, retrieval, transfer, logs and supporting services with explicit quantities/units. Compare alternatives at the same availability and recovery targets. Managed services can lower operating effort but are not automatically cheaper in every usage pattern.
S3 Intelligent-Tiering has no minimum storage duration. Eligible objects move among automatic tiers; Archive Access and Deep Archive Access are optional and require restore workflows. Objects smaller than 128 KB remain Frequent Access without monitoring/automation fees. IA and Glacier classes have their own minimum durations and retrieval/minimum-size charges.
Model each network hop, including NAT/PrivateLink/TGW and transfer exemptions. S3/DynamoDB gateway endpoints have no endpoint processing/hour fee; interface endpoints differ. S3-to-CloudFront origin transfer has applicable exemptions. Do not compromise AZ resilience merely to eliminate transfer.
Storage tiering revision matrix
| Class | Access/restore model | Minimum storage-duration charge |
|---|---|---|
| Standard | Immediate supported access | None |
| Intelligent-Tiering | Automatic instant tiers; optional restore tiers | None |
| Standard-IA / One Zone-IA | Immediate, with retrieval charges | 30 days |
| Glacier Instant Retrieval | Immediate, with retrieval charges | 90 days |
| Glacier Flexible Retrieval | Restore options | 90 days |
| Glacier Deep Archive | Longer restore options | 180 days |
Minimum billable object size and metadata are separate from duration. Intelligent-Tiering’s optional archive tiers must be enabled; a 30-day IA rule from another class cannot be assigned to the whole storage class.
For a hypothetical us-east-1 calculation dated 2026-10-12, assume 1,000 GB storage at $0.02/GB-month and 300 GB of retrieval at $0.01/GB: storage=$20 and retrieval=$3 before requests, transitions and delivery. Compare a candidate with a cheaper storage rate against the same traffic/retention model; cost per GB stored alone is insufficient.
Track data transfer by source/destination and intermediate NAT/TGW/PrivateLink. Ingress/egress and same-AZ/Region exceptions are service-specific. An endpoint may remove NAT processing without removing NAT’s hourly cost if other traffic still requires it. Include recurring and one-time migration costs in the calculator estimate.
Scenario and rationale
A new document portal stores 20 TB, with 80% rarely read but legally required to be retrievable within one minute. Deep Archive is inappropriate for those documents despite cheap storage. Compare Standard-IA, Intelligent-Tiering automatic instant-access tiers and Glacier Instant Retrieval using access frequency, object size and retention.
Reject the cheapest archive row without restore-latency analysis. Reject a three-year compute commitment before uncertain launch traffic stabilizes. Use On-Demand initially and reassess the measured hourly floor after rightsizing.
Practice drill
Prepare two calculator estimates using a supplied storage-size/access distribution. Include retrieval requests, bytes, transition charges, replication and end-user delivery. Sensitivity-test a tenfold retrieval month and an early deletion event. Record owner budgets and anomaly alerts as delayed safeguards, not a hard cap.
References
- Official Domain 2 task guide
- Intelligent-Tiering rules
- S3 pricing dimensions
- Well-Architected cost modeling
Return to Domain 2. The drill is proposed study work, not a claim of a completed deployment.