FinOps Certified Engineer flashcards

32 cards. Read the term and the definition here. The deck at the top of this page is optional.

Request-based allocation

Pod cost computed from container CPU and memory requests multiplied by runtime, then mapped to node price. Simple and stable, but overstates cost when requests greatly exceed real usage.

Usage-based allocation

Pod cost derived from measured consumption (actual CPU and memory used) rather than requests. More accurate for over-provisioned clusters but requires reliable metrics pipelines and sampling.

Idle container capacity

Node cost not consumed by any pod allocation, covering unallocated requests and unused node resources. It must be attributed explicitly to a shared or platform cost center.

Bin packing efficiency

Ratio of total pod requests to total node allocatable capacity in a cluster. Low values indicate fragmentation and wasted node spend, often fixed by rightsizing and instance diversity.

Namespace cost allocation

Mapping cluster spend to namespaces as the primary ownership boundary, then splitting shared overhead such as system pods, monitoring, and ingress by an agreed rule.

Label-based cost allocation

Using Kubernetes labels such as team, app, environment, and cost-center as the allocation key. Missing or inconsistent labels create untaggable spend, so policy enforcement is a cost control.

Cluster autoscaler scale-down

Behavior that removes underutilized nodes when pods can be rescheduled elsewhere. Scale-down thresholds and pod disruption budgets determine how quickly savings appear.

Horizontal Pod Autoscaler

Controller that scales replica count against metrics such as CPU utilization relative to requests. Changing requests alters its scaling behavior and therefore cost.

Sidecar cost overhead

Spend attributable to injected helper containers such as service mesh proxies and log shippers. Measured per namespace and usually allocated as shared platform overhead.

Spot node pools

Node capacity bought at a discount with interruption risk. Suitable only for interruption-tolerant workloads such as stateless services and queue-driven batch jobs.

Pod disruption budget

Policy limiting how many pods of a workload may be evicted simultaneously. It constrains consolidation and scale-down, directly affecting how fast node savings are realized.

Vertical pod rightsizing

Adjusting container requests and limits to match observed usage percentiles. Lowering requests reduces reserved capacity and node count but risks throttling or eviction if set too low.

Shared cluster overhead

Cost of system components, monitoring, ingress, and control plane that serve all workloads. Allocated by namespace share or treated as a platform chargeback line.

Resource quota governance

Namespace-level caps on requests and limits that prevent unbounded reservation. Quotas bound worst-case spend and force teams to justify capacity growth.

Overcommit ratio

Ratio of total pod limits to allocatable node capacity. High overcommit improves packing but raises contention and eviction risk, so it is a deliberate cost-versus-reliability tradeoff.

Amortized cost

Commitment upfront fees spread evenly across the term, so daily cost reflects the benefit period. Contrasts with unblended cost, which books the whole charge on the purchase date.

Effective savings rate

On-demand equivalent cost minus actual cost, divided by on-demand equivalent cost. Reported per commitment family and scope rather than as one blended figure.

Commitment coverage

Share of eligible spend covered by commitments. Low coverage signals exposure to on-demand rates; it is assessed together with utilization to avoid over-purchasing.

Commitment utilization

Portion of purchased commitment actually consumed during the period. Low utilization means paid-for discount is wasted and signals over-purchase or workload migration.

Inform phase

FinOps phase delivering visibility, allocation, and benchmarking. Engineering outputs include tagged resources, cost allocation models, and dashboards that owners trust.

Optimize phase

FinOps phase covering rightsizing, rate optimization through commitments, and usage elimination. Engineering work includes automation, architecture change, and waste removal.

Operate phase

FinOps phase establishing governance and automation: budgets, anomaly response, policy enforcement, and recurring review cadence with named owners.

Unit economics metric

Cost divided by a business volume such as requests, tenants, or transactions. Normalizes spend against growth and shows whether efficiency is improving over time.

Showback reporting

Distributing cost reports to resource owners without financial consequence. Requires the same allocation and tagging integrity as chargeback but carries no budget impact.

Chargeback model

Posting allocated cost to team or business-unit budgets. Demands defensible allocation rules, dispute handling, and reconciliation to the cloud invoice.

Anomaly detection baseline

Expected spend model per account or service incorporating seasonality. Alerts fire on deviation; each alert needs an owner and runbook to avoid fatigue.

Tagging enforcement policy

Mandatory tags applied at provisioning through infrastructure-as-code and cloud policy. Untagged resources are blocked or queued for remediation and reported as exceptions.

Rightsizing lookback window

Period of usage data used to size resources, typically covering weekly and monthly cycles. Longer windows capture peaks but slow response to changing demand.

Forecast variance

Difference between forecast and actual spend, tracked by service and period. Persistent bias indicates a broken driver model rather than random error.

Cost allocation model

Documented rules mapping shared and direct cost to owners. Must be reproducible, reconciled to the invoice, and reviewed when architecture or org structure changes.

Budget alert threshold

Percentage of budget that triggers notification, often staged at fifty, eighty, and one hundred percent. Thresholds should map to actions, not just messages.

Discount sharing scope

Organizational boundary within which commitment discounts are redistributed, such as a payer account. Scope choice changes which teams see savings and how incentives align.