Cloud cost modeling cannot be an afterthought
When cost modeling arrives at the end of a modernization assessment, engineering leaders react to numbers instead of designing to them.
T-shirt-size cloud cost estimates from consulting engagements don't survive a FinOps review — because they are not grounded in actual workload behavior and sizing data.
When cost models are produced separately from the roadmap, wave sequencing cannot optimize for cost — it optimizes for technical convenience and ignores the economics.
An executive sponsor cannot defend a multi-year modernization program to the board with a cost estimate expressed as a range. The model needs to be specific, workload-level, and traceable.
Generate workload-level AWS cost models from observed resource consumption. saasups measures actual CPU, memory, storage, and network consumption from the estate map and maps each workload to specific AWS service configurations — EC2 instance types, RDS configurations, managed service equivalents — producing a cost model grounded in real behavior, not T-shirt sizing.
Embed cost modeling in every wave of the roadmap. FinOps outputs are produced per migration wave — not as a separate deliverable — so the sequencing can account for cost optimization opportunities such as Reserved Instance commitment timing and managed service substitution sequencing.
Deliver a before-and-after cost model the board can defend. saasups produces a current-state infrastructure cost baseline and an AWS target-state cost projection per workload, per wave, and in aggregate — with the assumptions documented so the model can be audited, not just presented.
FinOps cost modeling embedded in the modernization roadmap
Observed resource consumption baseline
saasups measures actual CPU, memory, storage, and network consumption for every workload in the estate — producing a current-state cost baseline from real behavior, not assumptions.
AWS service mapping and rightsizing
Each workload is mapped to specific AWS service configurations — EC2 instance family and size, RDS engine and class, managed service equivalents — based on observed resource consumption and the recommended modernization strategy.
Reserved Instance and Savings Plan modeling
saasups identifies Reserved Instance and Savings Plan commitment opportunities per migration wave and models the cost impact of commitment timing alongside the migration sequence.
Total cost of migration per wave
Migration effort cost — compute, data transfer, testing, and rollback reserve — is estimated per wave alongside the steady-state AWS run cost, so the full economics of each wave are visible before commitment.
How saasups FinOps cost modeling works
saasups derives the current-state cost baseline from observed resource consumption in the estate map. Actual CPU utilization, memory footprint, storage consumption, and network throughput for each workload are measured from runtime behavior — producing a per-workload cost baseline that reflects what the system is actually consuming, not what it was provisioned to consume.
The baseline feeds the AWS cost projection engine. Each workload's recommended modernization strategy determines the target AWS service set. saasups maps observed resource consumption to specific AWS service configurations — EC2 instance family and size, RDS class and engine, ECS or EKS task sizing — and prices them against current AWS rate cards with configurable region and commitment assumptions.
Cost models are produced at three levels: per workload, per migration wave, and in aggregate. Each level includes the current-state baseline, the AWS target-state projection, the migration effort estimate, and the payback period. Reserved Instance and Savings Plan opportunity is identified per wave, and commitment timing is modeled against the migration sequence so financial commitments align with workload availability.
Frequently asked questions
saasups models are derived from observed resource consumption, not from T-shirt sizing assumptions. Accuracy improves with the quality of the estate ingestion — environments with 30 days or more of observed behavior produce the most representative cost baseline. All assumptions are documented in the output.