A billing discount applied to specific instance types in a specific AZ in exchange for a 1-yr or 3-yr commitment. Provides up to 75% savings vs on-demand. Best for stable, predictable production workloads.
AWS
SP
Savings Plan
A flexible billing model: commit to a dollar amount of hourly compute spend and receive discounted rates across instance families, OS types, and AZs. Up to 72% savings. Lower risk than RIs due to broader flexibility.
AWS
OD
On-Demand
Pay per second/minute with no commitment. No upfront cost. Highest price but maximum flexibility. Use for spike traffic, testing, or workloads you cannot predict.
baseline
Spot
Spot Instance
Spare EC2 capacity sold at 60–91% discounts. Can be interrupted by AWS with 2-minute warning. Use for batch processing, CI/CD, ML training with checkpoints. Do NOT use for anything requiring uptime guarantees.
AWS
Azure Spot
Azure Spot VM
Azure equivalent of AWS Spot Instances. Up to 90% off standard VM pricing. Interruption occurs when Azure needs capacity back. Configure eviction policies and checkpointing.
Azure
GCP Spot
GCP Spot / Preemptible VM
GCP's discounted compute option. Up to 91% off on-demand pricing. VMs can be preempted at any time. Pre-empted after a maximum of 24 hours. Best for fault-tolerant batch workloads.
GCP
Coverage
RI / SP Coverage
Percentage of your compute spend that is covered by Reserved Instances or Savings Plans. Target: 70–85% for baseline workloads. Below 50% means paying on-demand premiums unnecessarily.
metric
OD Gap
On-Demand Gap
The portion of compute spend not covered by any commitment (RI, SP, or Spot). The goal is to minimize this while avoiding over-commitment on variable workloads.
metric
// B. Compute Optimization
RS
Right-Sizing
Matching instance specifications (CPU, RAM) to actual workload requirements. If average CPU is below 40% over 30 days, the instance is oversized. Typical savings: 30–60% per instance after downsizing.
high ROI
Downsize
Downsizing
The act of replacing an oversized instance with a smaller one (e.g., r5.xlarge → r5.large). Always test at the target size before terminating the original. Set alerts before closing the old instance.
action
Idle
Idle / Orphaned Resource
A running instance or attached storage volume with no active workload. Common sources: stopped containers with persistent volumes, test instances left running, EBS volumes attached to terminated instances. Average waste: 10–15% of compute spend.
waste
CO
Compute Optimizer
AWS service (also Azure Advisor, GCP Recommender) that analyzes utilization patterns and recommends right-sizing opportunities. AWS Compute Optimizer covers EC2, ASG, Lambda, and EBS volume types.
AWS Tool
ASG
Auto Scaling Group
A set of EC2 instances managed together with automatic scaling based on metrics (CPU, memory, request count). Right-configured ASGs prevent over-provisioning during low-traffic periods.
AWS
Sched.
Scheduled Shutdown
Automating the stop/start of non-production instances on a schedule. Dev/staging environments running 24/7 but used 9–5 can save ~65% on compute by stopping outside business hours.
automation
// C. Storage & Data Tiering
Hot
Hot / Standard Storage Tier
The default, highest-performance storage tier. For frequently accessed data (daily–weekly). Highest storage cost but zero retrieval penalty. Do not use for data accessed less than monthly.
baseline cost
Cool / IA
Cool / Infrequent Access
Storage tier for data accessed 1–12 times per month. Lower storage cost (40–60% savings vs Hot) but charges for retrieval. Min/max storage duration policies apply.
savings
Archive
Archive / Cold Storage Tier
For data accessed quarterly or less. 70–80% cheaper than Hot storage. Retrieval takes 12–48 hours. Best for compliance archives, old logs, quarterly reports.
savings
Glacier
Glacier / Deep Archive
Lowest-cost tier for data with retrieval time of hours to days. 90–95% cheaper than Hot. Use for legal retention, long-term backup, data that will never be accessed but must be retained.
lowest cost
LC
Lifecycle Policy
Automated rules that transition objects between storage tiers as they age. Example: Hot → Cool (90 days) → Glacier (365 days). Set at the bucket level — always use bucket-level policies, not per-object rules.
automation
EBS
EBS Volume & Snapshot Waste
Elastic Block Store volumes remain attached to terminated instances. Snapshots accumulate from failed automation runs. Audit monthly: aws ec2 describe-volumes --filters Name=status,Values=available
waste
// D. Tagging & Cost Allocation
Tag
Resource Tagging
Key-value metadata applied to cloud resources for cost allocation, ownership identification, and automation. Example tags: Environment=production, Owner=platform-eng, CostCenter=CC-12345.
foundation
CAT
Cost Allocation Tag
AWS feature that makes user-defined tags visible on cost reports. Tags must be activated in the Billing console to appear on cost allocation reports. Azure and GCP have equivalent features.
AWS
CB
Chargeback / Showback
Chargeback = direct billing to teams/departments. Showback = showing costs to teams without actual billing. FinOps typically starts with Showback to build cost awareness before moving to Chargeback.
process
CC
Cost Center
A finance category used to group and attribute costs. Each cloud project or team maps to a Cost Center code. Enables finance to see cloud spend mapped to organizational structure.
finance
Budget
Cost Budget
A defined spending limit with alert thresholds (e.g., 50%, 80%, 100%). Best practice: budgets per team/project, not just overall. Integrate anomaly alerts for >20% week-over-week spikes.
control
// E. Networking & Egress
Egress
Egress Cost
Charges for data leaving a cloud region or provider. Egress is one of the most overlooked cost categories. Cross-region transfers and third-party CDN origin pulls are common sources of unexpected egress charges.
common surprise
CDN
Content Delivery Network
A geographically distributed network of edge servers that caches content close to users. Reduces egress costs from origin servers by serving cached content from edge locations. CloudFront, Azure CDN, Cloudflare.
cost reducer
NAT GW
NAT Gateway
A managed service that allows private subnet instances to access the internet while preventing inbound connections. Charged per hour + per GB of data processed. For high-egress workloads, consider a NAT instance instead.
$0.045/hr+
[ AdSense Slot — mid Glossary ]
// F. FinOps Framework & Process
FinOps
Cloud Financial Management
The practice of bringing finance, engineering, and operations together to manage cloud costs. Three phases: Crawl (visibility) → Walk (optimization) → Run (automation). Defined by the FinOps Foundation.
framework
Waste
Cloud Waste
Cloud spend that produces no business value: idle resources, over-provisioned instances, lapsed commitments, data in wrong storage tiers. Industry average: 15–32% of total cloud spend.
15–32% avg
Anomaly
Cost Anomaly
An unexpected or unexplained change in cloud spend. Anomaly detection uses ML to identify unusual patterns. Best practice: alert on >20% week-over-week spend increase per service per account.
investigate
TCO
Total Cost of Ownership
The complete cost of running a workload, including compute, storage, networking, data transfer, licensing, and operational overhead. Use TCO models to compare cloud vs. on-premises vs. hybrid.
analysis
UE
Unit Economics / KPI
The cost per unit of business output: cost per API call, cost per user per month, cost per transacted dollar. FinOps Run stage teams track unit economics as the primary efficiency metric.
Disclaimer: This guide provides general informational content about cloud infrastructure cost management. Figures and benchmarks are based on publicly available industry averages (e.g., Gartner, IDC, cloud provider documentation) and may vary by provider, region, and workload. This content is not a substitute for professional financial, legal, or technical advice specific to your organization.