Put your next cloud-cost question to work.
Explain a bill, assess a resource or verify a change. Start with a practical guide; use the glossary below when you need a definition.
On this page
Start with a question
Practical guides 7 guides
What is FinOps?
A plain-English operating model for finance, engineering, and product teams.
Cloud cost optimization
How evidence, pricing, and ownership workflows reduce recurring cloud waste.
How to reduce Azure costs
A practical sequence for finding idle resources, right-sizing, and reporting.
Right-size Azure VMs
How to evaluate safer VM SKU changes with utilization and pricing evidence.
Azure Reserved Instances
When reservations save money and how to avoid over-committing capacity.
Azure net vs list price
Why savings estimates should reconcile to the customer's actual billed rate.
Azure cost optimization checklist
A field checklist for audits, evidence, pricing, ownership, and follow-up.
Azure field guide
Resource conditions, evidence limits and review steps.
Storage 3 guides
Unattached managed disk
An Azure managed disk remains allocated and billable but is not currently attached to a virtual machine. Unattached is a strong signal, not automatic permission to delete data.
Old snapshot
A snapshot is older than an organization’s documented recovery or retention window and appears to have no active legal, compliance, forensic, or recovery purpose. Age alone never proves a snapshot is safe to delete.
Hot-tier storage candidate
Blob data is a cooler-tier candidate when access patterns, retention needs, retrieval latency, minimum-duration rules, and total cost indicate Hot is no longer the right tier. Lower storage price alone is not the decision.
Compute 2 guides
Oversized virtual machine
A VM is a rightsizing candidate when observed demand and known workload constraints may fit a less expensive compatible SKU with safe headroom. No single CPU threshold proves a production VM is oversized.
Stopped-but-billing VM
A VM stopped from inside the guest can remain allocated. Compute billing generally stops only when Azure reports the VM as deallocated; attached disks, public IP configurations, and other resources may still bill after deallocation.
Networking 2 guides
Unused public IP
A public IP resource has no current Azure association that justifies its recurring charge. An empty direct association is useful evidence, but DNS, firewall, allowlist, IaC, or future failover dependencies can exist outside that relationship.
Idle load balancer
A load balancer may be idle when backend pools, health, flows, and workload context show it no longer serves a live path. Empty pools are stronger evidence than low traffic, but neither automatically proves deletion is safe.
PaaS 1 guides
Database 1 guides
Commitments 1 guides
Governance 1 guides
Glossary
FinOps
FinOps is an operational framework and cultural practice for maximizing the business value of technology. Engineering, finance and business teams use timely data to make decisions and establish financial accountability.
Adapted from the FinOps Foundation Framework, licensed under CC BY 4.0.
- Core Loop
- Inform → Optimize → Operate
Reserved Instances
A billing discount model where a customer commits to utilizing a specific amount of cloud compute or database capacity (typically a 1-year or 3-year term) in exchange for significantly reduced hourly rates compared to standard on-demand pricing.
- Savings Ratio
- Commitment Rate = On-Demand Rate × (1 - Discount %)
Net vs. List Price
List price represents the standard public retail rate advertised by cloud providers. Net price represents the actual billed rate after incorporating custom enterprise discounts, negotiated enterprise agreements (EA), or reservation models. Finding calculations must always use net rates to prevent double-discounting errors.
- Net Calculation
- Net Cost = List Retail Cost × (1 - Enterprise Discount %)
Idle Resource
An active cloud resource (such as a Virtual Machine, Database, or Load Balancer) that continues to run and accrue static hourly charges but performs little or no meaningful operational work. Typically identified by sustained low CPU, memory, IOPS, and network throughput thresholds over a 14-day to 30-day window.
- Idle Threshold
- Max CPU < 5% && Active IOPS = 0 over 30 Days
Cloud Cost Audit
A systematic, data-backed inspection of the cloud subscriptions, accounts or projects you authorize. The audit analyzes available performance telemetry, billing records, and resource metadata to produce reviewable savings and risk findings with source and limitation context.
- Audit Integrity
- Finding = Metric Evidence + Net SKU Ratio Match
Turn the checklist into a repeatable audit.
Costframe connects read-only, records the evidence behind each finding, and keeps list-price estimates separate from net savings.