In the SAA-C03 exam, cost optimization questions make up roughly 20% of the total. Storage cost optimization is one of the most common topics, with questions asking "how do you store the same data more cheaply?"
What Are S3 Storage Classes?
Think about how you organize documents. Papers you look at every day stay on your desk. Files you check occasionally go in a drawer. Old archives get boxed up in a storage room. Ancient records get moved to a long-term archive facility. S3 works the same way — by moving data to cheaper storage based on how often you access it, you can dramatically reduce costs.
S3 Storage Classes Overview
| Class | Availability | Retrieval Fee | Minimum Duration | Best For | |-------|-------------|---------------|-----------------|----------| | S3 Standard | 99.99% | None | None | Frequently accessed data | | S3 Standard-IA | 99.9% | Yes | 30 days | Less than once a month | | S3 One Zone-IA | 99.5% | Yes | 30 days | Reproducible data, single AZ | | S3 Intelligent-Tiering | 99.9% | None | None | Unpredictable access patterns | | S3 Glacier Instant Retrieval | 99.9% | Yes | 90 days | Quarterly access, millisecond retrieval | | S3 Glacier Flexible Retrieval | 99.99% | Yes | 90 days | Annual access, minutes to 12 hours | | S3 Glacier Deep Archive | 99.99% | Yes | 180 days | 7+ year compliance, 12-48 hours |
S3 Lifecycle Policies — Automated Savings
A lifecycle policy automatically moves data to cheaper storage classes as it ages. Like your desk documents moving to a drawer after a month, then to the storage room after six months, automatically.
A typical lifecycle for log data:
0–30 days: S3 Standard (frequent analysis access) 30–90 days: S3 Standard-IA (occasional queries) 90 days to 1 year: S3 Glacier Flexible Retrieval (rarely accessed) After 1 year: S3 Glacier Deep Archive (compliance archiving)
Lifecycle policies support two types of actions: Transition: move data to a different storage class Expiration: automatically delete data after a specified period
When to choose this: look for "automatically move aging data to cheaper storage" or "change storage classes over time automatically".
S3 Intelligent-Tiering — Let AWS Decide for You
When you cannot predict access patterns, use S3 Intelligent-Tiering. AWS monitors each object's access frequency and automatically moves it to a cheaper tier if untouched for 30 days, then instantly moves it back to the Standard tier when accessed again. No retrieval fees, but a small per-object monitoring fee applies.
When to choose this: look for "unpredictable access patterns" or "automatic storage optimization".
Glacier Archiving — Long-Term Storage Options
Glacier provides ultra-low-cost storage for data you rarely access. Three options differ in retrieval time and cost.
Glacier Instant Retrieval
Like having items in a warehouse where a single phone call gets them delivered immediately. Retrieval is in milliseconds, making it suitable for medical images, news archives, or other data accessed quarterly but needing fast retrieval when needed.
Retrieval time: milliseconds Minimum storage duration: 90 days Cheaper than S3 Standard-IA
Glacier Flexible Retrieval
Like shipping items from a warehouse by truck — not fast, but affordable. Choose between an expedited tier (1–5 minutes) for urgency or bulk retrieval (up to 12 hours) for free.
Retrieval time: 1 minute (expedited) to 5 hours (standard) to 12 hours (bulk, free) Minimum storage duration: 90 days Cheaper than Glacier Instant Retrieval
Glacier Deep Archive
The deepest warehouse — retrieval takes days but the cost is extremely low. Use it for data that regulations require you to keep for 7 years or more, such as financial or medical compliance records.
Retrieval time: 12 hours (standard) to 48 hours (bulk) Minimum storage duration: 180 days The cheapest storage option in all of AWS
When to choose this: look for "long-term compliance archiving", "cheapest possible storage", "7-year retention requirement".
EBS Cost Optimization — Reducing Disk Costs
Switching from gp2 to gp3
gp3 is approximately 20% cheaper than gp2 while allowing you to set IOPS and throughput independently of volume size. With gp2, IOPS scales with volume size. With gp3, you can specify exactly what you need without over-provisioning. In most cases, migrating existing gp2 volumes to gp3 saves money immediately.
Choosing the Right Volume Type
| Workload | Recommended Volume | Reason | |----------|-------------------|--------| | General web or app server | gp3 | Balanced cost and performance | | Log processing, big data | st1 (HDD) | Much cheaper than SSD, high throughput | | Infrequently accessed data | sc1 (HDD) | Cheapest HDD option | | High-performance databases | io2 | Guaranteed high IOPS (expensive, use only when needed) |
Cleaning Up Unused Resources
Regularly delete orphaned EBS volumes (not attached to any instance) Clean up old EBS snapshots (automate with Data Lifecycle Manager) When terminating EC2 instances, confirm root volume auto-delete is enabled
Hybrid Storage Cost Savings
AWS Storage Gateway — Connecting On-Premises to the Cloud
A hybrid service that connects on-premises storage to S3. Replacing physical storage infrastructure with S3 reduces hardware and maintenance costs.
File Gateway: access S3 from on-premises as if it were an NFS/SMB file server Volume Gateway: back up on-premises volumes to S3 as EBS snapshots Tape Gateway: replace physical tape backup devices with S3 and Glacier
When to choose Storage Gateway: look for "archive on-premises data to S3", "replace physical tapes with cloud", "access S3 from on-premises as NFS".
AWS DataSync — Bulk Data Transfer
Rapidly moves large amounts of data from on-premises file servers to S3, EFS, or FSx. After migration, decommissioning on-premises storage hardware reduces ongoing costs.
Exam Key Points
"Automatically move aging data to cheaper storage" -- S3 Lifecycle Policy
"Unpredictable access patterns, automatic optimization" -- S3 Intelligent-Tiering
"Quarterly access, millisecond retrieval, 90-day minimum" -- Glacier Instant Retrieval