cloud cost calculator
Methodology
Last updated May 31, 2026
How Stratos makes a fair cloud pricing comparison
Stratos is designed for an early architecture question: if the same virtual-machine workload ran on AWS, Azure, Google Cloud, or Oracle Cloud Infrastructure, what public list-price range should a team expect before tax, discounts, support, and surrounding services?
The calculator does not try to replace provider quote tools or billing exports. Instead, it keeps the comparison narrow enough to audit. It uses a shared workload shape, reads local public pricing files, chooses the closest priced provider meter, separates the visible cost components, and shows the assumptions beside the result.
Inputs normalized across providers
The primary workload dimensions are vCPU, memory, processor preference, operating system, monthly compute hours, region strategy, and storage quantity. Those inputs are intentionally plain because they map to the way teams normally start a sizing conversation.
For AWS, Azure, and Google Cloud, CPU is shown as vCPU. OCI commonly exposes compute as OCPU, so Stratos converts the requested vCPU amount using 1 OCPU = 2 vCPU. That conversion is shown in the interface so a user can see both the cross-cloud request and the OCI equivalent.
Memory is treated as provisioned GB of RAM. If a provider exposes exact flexible sizing, Stratos can use exact CPU and memory quantities. If the provider uses fixed VM families, Stratos selects a close instance shape and marks the confidence as closest rather than exact.
Compute pricing
Base compute uses on-demand or pay-as-you-go public list pricing. The calculator favors comparable general-purpose or selected processor-family meters over promotional, spot, reserved, or contract-only prices. This keeps the first comparison reproducible even when the user has not entered account-specific discount information.
Processor selection is a filter, not a promise that every provider has the same silicon generation in every region. If a requested processor is not represented in the loaded pricing file for one provider, that provider is marked unavailable for that scenario. This is better than inventing a substitute price and hiding the mismatch.
Windows pricing is separated when the loaded data exposes a real Windows meter or a computable Windows license component. Linux and Oracle Linux scenarios use the base Linux VM meter. BYOL scenarios remove the customer-owned license cost from the public-cloud estimate, but Stratos does not validate whether a customer is licensed or eligible for a provider's BYOL program.
Storage pricing
Block storage is modeled as provisioned GB-month. The calculator groups provider products into broad classes such as standard HDD, balanced SSD, and performance SSD. Where providers bill in fixed disk sizes or performance units, Stratos exposes the basis in the pricing table so the rounded or selected meter can be reviewed.
File storage and object storage are optional add-ons. They are kept out of the default total until enabled because many compute comparisons do not need them. Object storage can include stored GB, class selection, read requests, write requests, and retrieval GB when those meters are present in the price files.
Some storage products have operational costs that are difficult to normalize across four clouds, such as replication, lifecycle transitions, snapshot churn, provisioned throughput, or minimum retention. Those charges are intentionally excluded unless the calculator can show a comparable meter for every provider in the active comparison.
Regions and monthly hours
Region choice matters because public cloud prices can vary by geography. Stratos supports provider-specific region selection for exact codes, and a region-family mode for teams that only need a broad geography such as US East, Europe West, or Singapore/Southeast Asia. Region-family mode still shows the selected provider region because cloud footprints are not perfectly identical.
The default monthly usage is 730 hours, which represents an average month. A 744-hour option is included for users matching a 31-day provider calculator view. Custom monthly hours can model scheduled environments, proof-of-concept systems, or non-production workloads that are stopped outside business hours.
How price data is loaded
The static site reads local JSON price artifacts for AWS, Azure, Google Cloud, and OCI. Each artifact is generated from provider pricing sources, then served with the calculator. The footer shows available refresh metadata so users can judge whether a provider file is current enough for their decision.
When a pricing file is missing or a real meter cannot be found, Stratos marks that provider unavailable rather than filling the gap with a hand-entered default. That behavior can make a comparison look less complete, but it keeps the result honest and easier to audit.
What the estimate excludes
The default result excludes network transfer, NAT gateways, managed databases, Kubernetes worker nodes beyond the selected compute shape, monitoring, log ingestion, backup policies, third-party marketplace licenses, support plans, credits, taxes, currency conversion, and private discounts. The optional managed Kubernetes add-on includes only real cluster/control-plane hourly meters. Some excluded areas are partially discoverable in cloud price catalogs, but they often require workload-specific assumptions that can overwhelm a simple VM comparison.
Quote commitment modes in the estimate builder are labeled as illustrative. They do not model provider-specific reserved instances, savings plans, Azure reservations, Google committed-use discounts, enterprise agreements, promotional credits, or special migration incentives.
Review checklist before using a result
- Confirm that each provider has a priced result rather than an unavailable state.
- Open the selected meter details and verify region, instance family, operating system, and storage class.
- Switch monthly hours from 730 to the operating schedule you expect to run.
- Add storage usage only when it belongs in the same workload decision.
- Validate the leading provider in its native calculator before making a purchase decision.