Stratos Compute Cost Comparison

Scenario builder

Size one compute workload, compare each provider on equal terms

Totals include compute and OS/license only. Storage, network, databases, Kubernetes, and other services are intentionally out of the active comparison.

Scope Preset Region mode AWS region Azure region GCP region OCI egress region Usage hours

Selected compute hours are normalized across every provider. Use custom monthly hours for partial-month or scheduled resources, such as 200 h/mo. Use 744 when matching a 31-day provider calculator view, such as an OCI Cost Estimator screen showing 1 instance x 744 hrs/month; otherwise 730 keeps AWS, Azure, GCP, and OCI on the same average-month basis.

Compute Workload Processor OS image Custom vCPU Custom memory GB

OCI equivalent: 4 vCPU = 2 OCPU. Stratos uses 1 OCPU = 2 vCPU for OCI sizing.

OS images are limited to common cross-cloud families. Windows BYOL shows real provider compute cost only; customer license cost and eligibility are external.

Advanced pricing inputs

Load balancers Load balancer type Load balancer hours h/mo Object reads x 10k Object writes x 10k Object retrieval GB

Advanced inputs are limited to dimensions that currently have comparable real pricing across AWS, Azure, GCP, and OCI. Provider-specific or partial-provider meters are kept out of the customer-facing comparison until all four providers can be matched.

GCP is unavailable until a real gcp-prices.json is generated. Run python3 fetch_cloud_prices.py --gcp --output-dir ..

Standardized capabilities

Compute-only comparison

Details

Compute: standardized by vCPU, memory, processor preference, OS family, license mode, and monthly hours.
OCI sizing: requested vCPU is converted with 1 OCPU = 2 vCPU.
Excluded: block storage, object storage, file storage, DNS, load balancers, Kubernetes, internet egress, databases, monitoring, and other cloud services.

Service catalog

Comparable services Stratos can capture

Details

Active rows are already wired into the calculator. Planned rows are good four-provider candidates, but stay out of the controls until real meters and tests are in place.

Active

Block Storage

AWS: EBS
Azure: Managed Disks
GCP: Persistent Disk
OCI: Block Volume

Compared by storage class and GB-month. Provider-specific performance knobs stay hidden until they can be normalized across all four clouds.

Active

Compute

AWS: EC2
Azure: Virtual Machines
GCP: Compute Engine
OCI: Compute VM

Compared by vCPU, memory, OS image, license mode, and monthly hours. OCI sizing uses 1 OCPU = 2 vCPU.

Active

DNS

AWS: Route 53
Azure: Azure DNS
GCP: Cloud DNS
OCI: OCI DNS

Compared by public DNS query volume. Hosted zones, DNSSEC, resolver endpoints, and traffic steering are not active until comparable real meters exist.

Active

File Storage

AWS: EFS
Azure: Azure Files
GCP: Filestore
OCI: File Storage

Compared by shared file storage GB-month. Performance units, operation fees, replication, and backup remain hidden until all four providers expose a normalized path.

Active

Internet Egress

AWS: Data Transfer Out
Azure: Bandwidth
GCP: Networking Egress
OCI: Outbound Data Transfer

Compared by outbound GB. OCI's free monthly allowance is applied before its regional egress meter.

Active

Load Balancer

AWS: ELB
Azure: Load Balancer / Application Gateway
GCP: Cloud Load Balancing
OCI: Network / Flexible Load Balancer

Compared by L4 or L7 load balancer count and hours. Processed data, rules, and usage units remain hidden.

Active

Managed Kubernetes

AWS: EKS
Azure: AKS Standard
GCP: GKE
OCI: OKE Enhanced Cluster

Compared by managed cluster/control-plane hours. Worker nodes, load balancers, storage, and egress are priced separately.

Active

Object Storage

AWS: S3
Azure: Blob Storage
GCP: Cloud Storage
OCI: Object Storage

Compared by Standard/Hot, Infrequent/Cool, Archive, stored GB, requests, and retrieval where real meters exist.

Blocked

Container Registry

AWS: ECR
Azure: Container Registry
GCP: Artifact Registry
OCI: Container Registry

Storage meters are captured for AWS, Azure, and GCP when available. The calculator control remains hidden because the loaded OCI public price list does not expose a comparable Container Registry storage meter.

Why blocked?

Captured: AWS ECR image storage, Azure Container Registry stored data, and GCP Artifact Registry storage.
Missing: OCI Container Registry storage is not exposed as a real meter in the loaded OCI public price list.
Activation rule: expose the calculator control only after all four providers have a comparable real stored-image GB-month meter.

Discovery

Managed MySQL

AWS: RDS MySQL
Azure: Azure Database for MySQL
GCP: Cloud SQL for MySQL
OCI: MySQL HeatWave

Discovery meters are captured for all four providers where available. The calculator control remains hidden until engine, compute, storage, backup, and HA assumptions are normalized without losing service differences.

Why not active yet?

Captured: AWS RDS MySQL instance/storage, Azure Database for MySQL Flexible Server vCore/storage, GCP Cloud SQL for MySQL vCPU/RAM/storage, and OCI MySQL HeatWave ECPU/storage/backup.
Not 1:1: providers bill managed database compute as instance-hours, vCPU/RAM, ECPU, storage tiers, backup storage, and optional HA or HeatWave capacity.
Activation rule: expose calculator controls only after the four-provider sizing model clearly separates compute, storage, backup, HA, and optional analytics acceleration.

Blocked

Managed PostgreSQL

AWS: RDS PostgreSQL
Azure: Azure Database for PostgreSQL
GCP: Cloud SQL for PostgreSQL
OCI: PostgreSQL Database

Discovery meters are captured for AWS, Azure, and GCP when available. The calculator control remains hidden because the loaded OCI public price list does not expose a comparable PostgreSQL Database meter.

Why blocked?

Captured: AWS RDS PostgreSQL instance/storage, Azure Database for PostgreSQL Flexible Server vCore/storage, and GCP Cloud SQL for PostgreSQL vCPU/RAM/storage.
Missing: OCI PostgreSQL Database is not exposed as a real meter in the loaded OCI public price list.
Activation rule: expose calculator controls only after all four providers have a comparable real engine, compute, storage, and HA model.

Planned

Monitoring And Logs

AWS: CloudWatch
Azure: Azure Monitor
GCP: Cloud Monitoring / Logging
OCI: Monitoring / Logging

Candidate for log ingest, log retention, metrics, alarms, and API calls, with high-friction assumptions because billing models differ.

Blocked

NAT Gateway

AWS: NAT Gateway
Azure: NAT Gateway
GCP: Cloud NAT
OCI: Not exposed in loaded price list

Captured for AWS, Azure, and GCP, but not shown as a calculator control until OCI has a real comparable service meter.

Planned

Secrets And Keys

AWS: Secrets Manager / KMS
Azure: Key Vault
GCP: Secret Manager / Cloud KMS
OCI: Vault

Candidate for secret count, key count, and operation pricing after storage and request meters are separated cleanly.

Planned

Snapshots And Backup

AWS: EBS Snapshots / AWS Backup
Azure: Disk Snapshots / Azure Backup
GCP: Persistent Disk Snapshots
OCI: Block Volume Backup

Real meters are partially captured today, but the visible control stays deferred until the four-provider mapping is cleaner.

Pricing results

Monthly comparison

All visible prices are USD, on-demand/PAYGO, and pre-tax. BYOL scenarios exclude customer-owned license costs. Support, private discounts, credits, and unpriced third-party charges are not included.

Monthly estimate

Cost breakdown

Compute OS/license

Selected meters

Side-by-side SKU transparency

Details

Pricing inputs

Compute data sources

Details

Provider	Region	Confidence	Compute basis	Cost split	Pricing inputs

Pricing assumptions Comparable, not identical Details

Stratos now compares compute resources only. It maps each workload to the closest available on-demand VM compute meter from each provider, separates Windows OS/license cost when real pricing exposes it, and shows the exact SKU or meter used. Custom workloads use the entered vCPU and memory amounts for AWS, Azure, and GCP; OCI sizing converts the same vCPU request using 1 OCPU = 2 vCPU.

All totals are shown in USD before tax. Public list pricing can differ from invoice pricing because of credits, private discounts, support plans, taxes, and currency conversion.
Only compute and OS/license cost lines are included in totals.
Storage, networking, DNS, load balancers, Kubernetes, managed databases, backups, observability, support, and private discounts are excluded.
Compute hours are normalized across every provider. The default is 730 h/mo; custom monthly hours can model scheduled resources such as 200 h/mo, and selecting 744 h/mo applies the same 31-day month basis to AWS, Azure, GCP, and OCI.
Compute uses on-demand or PAYGO pricing only, with no reserved, spot, committed-use, savings plan, or enterprise discounts.
CPU sizing is entered as vCPU for AWS, Azure, and GCP; OCI OCPU is calculated from that request as 1 OCPU = 2 vCPU.
Presets use resources that are priced across the current comparison data.
Processor and OS selectors are matched against loaded provider data. If a selected processor, OS image, or license model has no real meter for a provider, that provider is marked unavailable.
Providers without a real pricing JSON file are shown as unavailable rather than estimated from defaults.
Region-family mode maps each provider to the closest comparable public region, but exact region codes are still shown because clouds do not always share a perfect regional footprint.
Taxes, support, and unpriced third-party license costs are excluded.
The OS selector is intentionally limited to common cross-cloud image families: Ubuntu/Oracle Linux and Windows Server. Provider-specific images such as Autonomous Linux are excluded from the main comparison.
Ubuntu/Oracle Linux uses the provider base Linux VM meter. Windows PAYGO adds a real Windows license meter only when the pricing file includes one. Windows BYOL is a customer-license scenario: Stratos shows real provider compute cost and excludes the external license cost, but it does not validate your licensing entitlement.