Start 14-day free trial

AWS Storage Cost Optimization: The SRE Playbook for gp3 and S3 Lifecycle Engineering

Slash AWS storage costs by 35%. Master gp2 to gp3 migration math, S3 lifecycle policies, and unattached disk cleanup with CostGlass flat-rate FinOps.

AWS Storage Cost Optimization: The SRE Playbook for gp3 and S3 Lifecycle Engineering
Figure 1: Architectural overview of AWS Storage Cost Optimization: The SRE Playbook for gp3 and S3 Lifecycle Engineering.

Migrating legacy Elastic Block Store (EBS) gp2 volumes to gp3 and implementing precise Simple Storage Service (S3) lifecycle transitions can instantly reduce storage expenses by 20% to 50%. Yet, because storage resources accumulate incrementally across hundreds of developer accounts, these opportunities are frequently overlooked until they comprise a major portion of the cloud bill.

This guide outlines the precise technical mechanics of EBS and S3 optimization, provides concrete remediation steps, and explains how to translate engineering actions into board-ready gross margin impact.

The Real Cost of Storage Inertia

Unlike compute instances, which can be scheduled or shut down, storage is persistent. It incurs costs 24 hours a day, 365 days a year, regardless of whether the attached virtual machines are running or idle.

In many mature AWS environments, three primary architectural anti-patterns drive storage waste:

  1. Over-Provisioned gp2 Volumes: Relying on legacy gp2 volumes where Input/Output Operations Per Second (IOPS) and throughput are artificially tied to disk size, forcing teams to purchase excess gigabytes just to achieve performance targets.
  2. Orphaned EBS Volumes and Snapshots: Retaining block storage volumes after their host EC2 instances are terminated, or keeping daily backup snapshots indefinitely without a rotation policy.
  3. S3 Storage Class Mismanagement: Storing petabytes of historical logs, raw database backups, or transient analytical assets in S3 Standard forever, rather than tiering them to Glacier or Intelligent-Tiering.

By systematically addressing these areas, platform engineers can reclaim thousands of dollars in wasted monthly cloud spend.


EBS gp2 vs. gp3: Performance and Financial Math

In 2020, AWS introduced gp3 to decouple storage performance from capacity. Under gp2, volumes receive a baseline performance of 3 IOPS per GB, up to a maximum of 16,000. If an application requires 3,000 IOPS, engineers are forced to provision a 1,000 GB volume—even if they only need 50 GB of actual storage capacity.

In contrast, gp3 offers a flat baseline of 3,000 IOPS and 125 MB/s of throughput included in the base storage price, completely independent of volume size. You can provision a 10 GB volume and still receive 3,000 IOPS.

The Unit Cost Comparison

Metric gp2 Volume (Baseline) gp3 Volume (Optimized) Net Savings
Storage Cost per GB-month $0.10 $0.08 20% reduction
Baseline IOPS Included 3 IOPS per GB (min 100, max 16k) 3,000 (regardless of volume size) Free performance
Baseline Throughput Included N/A (tied to volume size) 125 MB/s (regardless of size) Free performance
Cost for 10 TB (3,000 IOPS) $1,000.00 $800.00 $200.00 (20% saved)
Cost for 100 GB (10,000 IOPS) $1,000.00 (forced to 3,333 GB size) $138.00 ($8.00 storage + $130.00 IOPS) $862.00 (86.2% saved)

Executing gp2 to gp3 Migrations Online

EBS volumes can be modified on-the-fly without detaching the volume or restarting the instance. AWS Elastic Volumes handles the transition in the background with zero downtime. SREs can execute this transition via the AWS CLI:

aws ec2 modify-volume \
  --volume-id vol-0123456789abcdef0 \
  --volume-type gp3 \
  --iops 3000 \
  --throughput 125

To perform this migration safely at scale, platform teams should use Terraform. This ensures your Infrastructure as Code (IaC) remains synchronized with your actual cloud environment, preventing configuration drift:

resource "aws_instance" "app_server" {
  ami           = "ami-0c55b159cbfafe1f0"
  instance_type = "m7g.xlarge"

  root_block_device {
    volume_type           = "gp3"
    volume_size           = 100
    iops                  = 3000
    throughput            = 125
    delete_on_termination = true
  }
}

S3 Lifecycle & Intelligent-Tiering Engineering

Amazon S3 is deceptively inexpensive at $0.023 per GB-month for the Standard tier. However, when storage grows to hundreds of terabytes, S3 bills become a significant component of your Cost of Goods Sold (COGS).

Understanding S3 Storage Tiers

  • S3 Standard ($0.023/GB): Designed for active, frequently accessed data.
  • S3 Standard-IA ($0.0125/GB): For data accessed less than once a month, but requiring millisecond retrieval. Beware of the retrieval fee ($0.01 per GB).
  • S3 Glacier Instant Retrieval ($0.004/GB): Ideal for medical records or media archives accessed once or twice a year, with instant retrieval.
  • S3 Glacier Deep Archive ($0.00099/GB): The lowest cost tier, designed for multi-year regulatory compliance backups with 12-hour retrieval times.

The S3 Transition Fee Trap

SREs often write aggressive lifecycle policies to transition objects to Glacier without calculating transition costs. AWS charges a transition request fee per 1,000 objects ($0.03 to $0.05 depending on the destination tier).

If you have 10 million small files (each under 100 KB) in an S3 bucket, migrating them to Glacier Deep Archive costs $500 in API requests alone, while the monthly storage savings might only be $5.

Rule of Thumb: Only transition large objects (> 1 MB) to cold tiers, or aggregate small files into larger tarballs before upload.

Implementing S3 Intelligent-Tiering

For buckets with unpredictable access patterns, S3 Intelligent-Tiering automatically moves objects between frequent, infrequent, and archive access tiers based on usage history. There are no retrieval fees, only a small monitoring fee of $0.0025 per 1,000 objects (for objects larger than 128 KB).

Here is an AWS CLI command to apply an Intelligent-Tiering configuration to an existing S3 bucket:

aws s3api put-bucket-lifecycle-configuration \
  --bucket my-production-data-lake \
  --lifecycle-configuration '{
    "Rules": [
      {
        "ID": "AutoTieringRule",
        "Status": "Enabled",
        "Filter": {
          "Prefix": ""
        },
        "Transitions": [
          {
            "Days": 0,
            "StorageClass": "INTELLIGENT_TIERING"
          }
        ]
      }
    ]
  }'

The Build vs. Buy Storage Governance Trap

To keep storage costs under control, many engineering organizations attempt to build in-house automation pipelines. Typically, this involves writing Python scripts that run as nightly AWS Lambda functions to query the EC2 API, locate unattached EBS volumes, find old snapshots, and alert engineers via Slack.

This DIY approach almost always fails over time:

  • Maintenance Overhead: API schemas change, scripts break, and SREs spend valuable engineering hours maintaining internal dashboards rather than focusing on core product features.
  • Alert Fatigue: Raw Slack alerts listing hundreds of unattached disks are ignored by busy development teams, leaving waste unaddressed.
  • No Cross-Cloud Unified Visibility: If your organization also uses Azure Managed Disks or Google Cloud Persistent Disks, SREs must build and maintain separate, complex scripts for each cloud provider.

How CostGlass Eliminates Storage Waste Out-of-the-Box

Instead of wasting internal engineering resources building fragile tracking tools, CostGlass connects to your AWS, Azure, and Google Cloud environments in five minutes using secure, read-only IAM roles.

CostGlass immediately identifies and aggregates storage waste across your entire multi-cloud footprint:

  1. Idle Volume Detection: Identifies unattached EBS volumes, Azure managed disks, and GCP persistent disks, highlighting exactly how much money is being wasted on orphaned storage.
  2. gp2 to gp3 Rightsizing Recommendations: Scans all active EBS volumes, calculates the potential savings of migrating to gp3, and provides the exact Terraform or AWS CLI commands to execute the migration.
  3. S3 Bucket Analysis: Visualizes storage growth trends and identifies buckets that are prime candidates for S3 Intelligent-Tiering or Glacier transition policies.

Furthermore, on the Scale plan ($1,499/mo), you do not even have to write the remediation code yourself. CostGlass includes 8 hours of dedicated senior DevOps and SRE engineer time every single month. Our engineers will write the Terraform pull requests, configure the S3 lifecycle policies, and clean up the orphaned volumes for you—delivering immediate ROI without distracting your core product team.


Step-by-Step Storage Remediation Workflow with CostGlass

To systematically eliminate storage waste inside your organization, follow this three-step workflow using CostGlass:

Step 1: Isolate Storage Spend in Cost Explorer

Log into your CostGlass account and navigate to the Cost Explorer. Filter your multi-cloud ledger by Service (e.g., AmazonEC2 and AmazonS3) and group by UsageType or Resource ID. This exposes the exact volumes and buckets driving your storage costs.

Step 2: Review Rightsizing Recommendations

Go to the Rightsizing Playbook to view a prioritized list of idle volumes and gp2 to gp3 migration candidates. CostGlass calculates the exact monthly savings for each resource, factoring in IOPS and throughput configurations to prevent performance degradation.

Step 3: Authorize Execution

If you are on the Scale or Enterprise plan, flag the approved recommendations in your dashboard. Your dedicated CostGlass SRE will pick up the tasks, write the necessary Terraform code, and coordinate with your team to safely apply the updates during your standard maintenance windows.


Frequently Asked Questions

Will migrating from gp2 to gp3 cause any application downtime?

No. AWS Elastic Volumes allows you to modify the volume type, size, and IOPS of active EBS volumes while they are attached to running instances with zero downtime. However, volume performance may be slightly degraded during the modification state, so we recommend executing migrations during off-peak hours.

Are there any scenarios where gp2 is cheaper than gp3?

No. At $0.08 per GB-month, gp3 storage is flatly 20% cheaper than gp2 ($0.10 per GB-month). The only edge case is if you provision extremely high IOPS (> 3,000) and throughput (> 125 MB/s) on very small volumes, where the add-on performance costs of gp3 could theoretically exceed the baseline cost. CostGlass's recommendation engine automatically detects these rare scenarios and only suggests migrations that guarantee positive savings.

How does S3 Intelligent-Tiering handle deep archiving?

S3 Intelligent-Tiering includes two opt-in archive access tiers: Archive Access (90 days of inactivity) and Deep Archive Access (180 days of inactivity). Once activated, objects are automatically moved to these ultra-low-cost tiers, matching the pricing of Glacier and Glacier Deep Archive, while maintaining automated retrieval if the data is ever requested.

Ready to stop paying the legacy storage tax? Learn how to automate storage cleanups and gp3 migrations by reviewing our Savings Playbook. To see how CostGlass's flat-rate plans compare to percentage-of-spend tools, check out our CostGlass vs. CloudZero Comparison. For transparent flat pricing starting at $0, visit CostGlass Pricing.

Ready to unify your multi-cloud costs?

See your AWS, Azure, GCP, and Kubernetes spend in canonical FOCUS 1.0 in under 15 minutes.

Start 14-day free trial →

More from the blog

View all →