The Cloud FinOps and DevOps Glossary
Every cloud bill hides a vocabulary. This glossary explains 156 cloud cost, FinOps, DevOps, Kubernetes, and AI infrastructure terms in clear, simple English, each with a mini blog, a real example, and FAQs.
A
AI Cost Management
Track and control what your AI and machine learning workloads really cost, from GPUs to tokens.
AI FinOps
FinOps discipline applied to AI workloads: GPUs, training, inference, and tokens. Also known as FinOps for AI or FinOps in AI.
AI Governance
The rules and guardrails that keep AI systems responsible, secure, and on budget.
AI Infrastructure
The compute, storage, and networking stack behind AI models, and usually the most expensive part of the bill.
Amortized Cost
Spread the price of a commitment evenly over time, so daily costs reflect reality.
Anomaly Detection
Catch unusual spend the day it happens, not when the invoice lands.
API Gateway
One managed front door for all your APIs, priced per request.
Auto-Tagging
Let automation label your cloud resources, because humans always miss some.
Autoscaling
Add capacity when demand rises, remove it when demand falls, automatically.
Availability Zone
An isolated data center within a cloud region, your unit of fault tolerance
AWS (Amazon Web Services)
The largest public cloud, and the bill most FinOps teams meet first.
AWS Cost Explorer
Amazon's built-in window into your AWS spend.
Azure (Microsoft Azure)
Microsoft's cloud platform, and the enterprise's most familiar on-ramp.
Azure Cost Management
Microsoft's native suite for watching and controlling Azure spend.
B
Bin Packing
Fit more workloads onto fewer machines, like packing a suitcase properly.
Blended Rate
One averaged price across accounts, convenient for finance, confusing for engineers.
Blue-Green Deployment
Run two copies of production and switch traffic between them for risk-free releases.
Budget Alerts
Automatic warnings before spend crosses the line, not after.
Budget Controls
Hard and soft limits that keep cloud spend inside the plan.
Burstable Instances
Cheap servers that idle quietly and sprint briefly, perfect for spiky small workloads.
C
Capacity Planning
Decide how much infrastructure you will need before you need it.
Chargeback
Teams do not just see their cloud costs, they actually pay them.
CI/CD (Continuous Integration / Continuous Delivery)
Automate the path from code change to production release.
Claude (Anthropic)
Anthropic's family of large language models, priced per token and common in enterprise stacks.
Cloud Billing
How cloud providers measure, price, and invoice everything you use.
Cloud Cost Allocation
Connect every dollar of cloud spend to the team, product, or customer that caused it.
Cloud Cost Governance
The rules, owners, and guardrails that keep cloud spend deliberate.
Cloud Cost Management
The end-to-end discipline of seeing, controlling, and optimizing cloud spend.
Cloud Cost Optimization
Cut what is wasted, keep what creates value, continuously.
Cloud Financial Management
Treat cloud spend as a financial discipline, not just a technical one.
Cloud FinOps
Finance and engineering managing cloud spend together, as one continuous practice.
Cloud Governance
The overall rulebook for how your organization uses the cloud.
Cloud Waste
Money spent on cloud resources that deliver no value: roughly a third of most bills.
Cluster
A group of machines working together as one system.
Commitment-Based Discounts
Promise the cloud provider steady usage, get a lower price in return.
Compliance
Prove your cloud meets the rules, from security standards to data laws.
Container
Package an application with everything it needs, so it runs the same everywhere.
Container Orchestration
Automate the running of containers at scale: placement, scaling, healing, networking.
Cost Allocation Tags
The specific tags whose only job is to route costs to owners.
Cost Anomaly
A spend pattern that breaks from normal, and the earlier you see it, the cheaper it is.
Cost Avoidance
The money you never spent, because you prevented the cost instead of cutting it.
Cost Center
The organizational bucket a cost belongs to in the company's books.
CSPM (Cloud Security Posture Management)
Continuous scanning that finds security misconfigurations before attackers do.
D
Data Egress Costs
The toll you pay when data leaves the cloud, often the bill's biggest surprise.
Databricks Cost Management
Keep the data and AI platform's consumption-based bill under control.
Dedicated Hosts
Rent the entire physical server, not just a slice of it.
DevOps
Development and operations as one team, one workflow, one responsibility.
DevSecOps
Security built into the delivery pipeline, not bolted on at the end
Docker
The tool that made containers mainstream.
Drift Detection
Spot when reality stops matching the infrastructure code that defined it.
E
Edge Computing
Run computing close to where data is created instead of in a distant central cloud.
Effective Savings Rate (ESR)
The one number that tells you how well your discount commitments actually perform.
Egress / Ingress
Data moving out of the cloud costs money. Data moving in is usually free. Plan for the difference.
Elasticity
The cloud's defining promise: capacity that grows and shrinks with demand.
Ephemeral Environments
Short-lived copies of your stack that exist only as long as they are needed.
Event-Driven Architecture
Systems that react to events as they happen instead of polling and waiting.
F
FinOps Foundation
The vendor-neutral body that standardizes how the industry practices FinOps.
FinOps Framework
The shared playbook for managing cloud spend: phases, principles, and practices.
FinOps Maturity Model
Crawl, walk, run: an honest way to grade and grow your FinOps practice.
FinOps Metrics
The numbers that tell you whether your cloud practice is actually improving.
FinOps Personas
FinOps is a team sport. The personas define who plays and what each player needs.
Forecasting (Cloud Cost Forecasting)
Predicting next month's cloud bill well enough to plan a business on it.
G
Gemini (Google)
Google's family of multimodal AI models, billed per token through the Gemini API and Vertex AI.
GitOps
Your Git repository becomes the single source of truth for what runs in production.
Golden Path
The well-paved, fully supported way to build and ship software inside your company.
Google Cloud Platform (GCP)
Google's cloud: strong in data, Kubernetes, and AI, with its own billing personality.
Governance Guardrails
Automated boundaries that prevent expensive or dangerous actions before they happen.
GPU Optimization
GPUs are the most expensive thing in your cloud. Optimizing them pays accordingly.
Granularity (Cost Granularity)
How finely you can slice cost data determines which questions it can answer.
H
High Availability
Designing systems that keep working when their parts fail.
Horizontal Pod Autoscaler (HPA)
Kubernetes' built-in mechanism for adding and removing pod copies as load changes.
Hybrid Cloud
One environment spanning your own data center and the public cloud, working as a whole.
Hyperscaler
The handful of cloud providers operating at planetary scale.
I
IaC (Infrastructure as Code)
Define infrastructure in files, not clicks, and get review, repeatability, and history for free.
IAM (Identity & Access Management)
Who can do what in your cloud. The control plane for both security and spend.
Idle Resources
Infrastructure that runs and bills while doing nothing useful.
Incident Management
The discipline of detecting, resolving, and learning from production failures.
Inference Cost
The ongoing cost of running a trained model in production. Training is the wedding; inference is the marriage.
Ingress Controller
The front door of a Kubernetes cluster, routing outside traffic to the right service.
Instance Family
Cloud servers come in families tuned for different jobs. Picking the right one is free money.
Internal Developer Platform (IDP)
A product your platform team builds so other engineers can ship without tickets.
J
K
Karpenter
The open-source autoscaler that picks the right nodes for Kubernetes, fast and cheap.
KPI (Key Performance Indicator)
The few numbers a team commits to moving, reviewed until they move.
Kubernetes
The operating system of the modern cloud: it runs your containers so you don't have to.
Kubernetes Cost Allocation
Splitting a shared cluster's bill into honest per-team, per-service costs.
L
Lift and Shift
Move applications to the cloud as they are, change them later, maybe.
LLM Pricing (Model Pricing)
How large language models are billed: per million tokens, in and out, at rates set by model tier.
LLMOps
The operational discipline for running large language models in production.
Load Balancer
The traffic director that spreads requests across servers so none of them drowns.
Logging
The written record of what your systems did, and one of the sneakiest line items on the bill.
M
Managed Services
Pay the cloud provider to run the database so your engineers can build the product.
Microservices
Split the application into small independent services that ship and scale separately.
MLOps
Getting machine learning out of notebooks and into reliable, repeatable production.
Monitoring
Watching known signals so you learn about problems before your customers do.
Multi-Cloud
Running across several cloud providers, by strategy or by accident.
Multi-Tenancy
Many customers, one shared platform, with isolation in software.
N
Namespace (Kubernetes)
Kubernetes' way of dividing one cluster into named, governable territories.
Node
The actual machine in a Kubernetes cluster. Nodes are what the cloud bills you for.
Node Pool
Groups of identical nodes, so one cluster can offer different hardware to different workloads.
NoOps
The ideal where operations is so automated that nobody operates anything by hand.
O
Observability
Being able to ask your systems new questions, not just check the old answers.
On-Demand Instances
Cloud's list price: maximum flexibility, zero commitment, highest hourly rate.
Orchestration
Coordinating many automated pieces into one reliable, repeatable whole.
Overprovisioning
Buying more capacity than the workload needs, then paying for the gap forever.
P
Platform Engineering
Building the internal product that makes every other engineering team faster and safer.
Pod
The smallest deployable unit in Kubernetes: your container, plus its resource claim.
Policy as Code
Rules that enforce themselves: governance written in code and run by machines.
Private Cloud
Cloud-style infrastructure operated for one organization alone.
Public Cloud
Shared, rented, elastic computing from providers like AWS, Azure, and GCP.
Q
R
RBAC (Role-Based Access Control)
Permissions assigned to roles, roles assigned to people. Simple, auditable, scalable.
Release Management
Getting software changes into production deliberately: planned, staged, reversible.
Reserved Instances (RI)
Commit to capacity for one or three years and pay up to 70 percent less for it.
Resource Tagging
Labels on every cloud resource that say what it is, whose it is, and why it exists.
Rightsizing
Matching every resource to what its workload actually needs. The classic cloud saving.
ROI (Cloud ROI)
What the cloud gives back for what it costs, measured rather than assumed.
S
Savings Plans
Commit to spending a fixed amount per hour and get discounts with built-in flexibility.
Secrets Management
Keeping passwords, keys, and tokens out of code and under control.
Security Posture Management
Continuously checking your cloud configuration against what secure should look like.
Service Level Objective (SLO)
A measurable reliability target that tells you when good enough is reached.
Service Mesh
A dedicated layer that handles how microservices talk: security, routing, and visibility.
Showback
Show every team exactly what its cloud usage costs, without moving any money.
Spot Instances
The cloud's spare capacity at up to 90 percent off, with the catch that it can vanish.
SRE (Site Reliability Engineering)
Software engineers applied to reliability: automate operations, measure everything, cap the toil.
Sustainability (Cloud Carbon)
Measuring and shrinking the carbon footprint of your cloud, usually by the same moves that shrink the bill.
T
Tagging Strategy
The agreed set of labels every cloud resource must carry so costs can be traced to owners.
Terraform
The most widely used tool for writing infrastructure as code across every major cloud.
Toil
Manual, repetitive operational work that scales with the system but adds no lasting value.
Tokenization
How AI models split text into billable chunks. Tokens are the meter on every LLM bill.
Total Cost of Ownership (TCO)
Everything a system really costs over its life, not just the price on the invoice.
U
Unallocated Costs
The slice of the cloud bill that cannot be traced to any team, product, or owner.
Underutilization
Paying for capacity that mostly sits idle: the default state of unmanaged cloud resources.
Unit Economics
Cloud cost expressed per unit of business value: per customer, per order, per request.
Usage-Based Pricing
Pay for what you consume: the pricing model that makes cloud flexible and bills variable.
V
Vendor Lock-In
When leaving a provider would cost more than staying: the price of convenience, compounded.
Vertical Pod Autoscaler (VPA)
The Kubernetes component that rightsizes each pod's CPU and memory requests automatically.
Virtual Machine (VM)
A complete computer running as software: the basic unit of cloud computing.
Virtual Private Cloud (VPC)
Your own isolated network inside a public cloud, with private addresses and controlled doors.
Volume Discounts
Prices that fall as usage rises: the cloud's built-in reward for scale.
X
Y
Z
#
Policies (Cloud Policies)
The written rules of the cloud: what is allowed, required, and forbidden.
Self-Service Infrastructure
Engineers provision what they need through paved paths, without filing tickets.
Virtual Tags
Cost labels applied in software after the fact, fixing what real tags missed.
Workload Optimization
Tuning a specific application and its resources so it does the same work for less.