Glossary
Cloud cost, in plain English
The terms behind cloud waste and FinOps — defined honestly, without the fluff.
AWS Cost Explorer
AWS's built-in tool for visualizing and querying historical cloud spend. It's great for seeing where the money goes — but it stops at visibility; acting on what it shows is a separate job.
AWS Graviton
AWS Arm-based processors (e.g. m6g, c7g) that deliver similar or better performance than comparable x86 instances at up to ~20% lower cost. Migrating compatible workloads is a common savings lever.
Bin-packing
Scheduling workloads so they fill each node efficiently instead of leaving many machines half-empty. Poor bin-packing is a major source of Kubernetes waste — you pay for node capacity that nothing is running on.
Cloud waste
Spend on cloud resources that deliver no value: over-provisioned instances, idle databases, unattached storage, forgotten environments. On a typical bill it runs 20–30% of the total.
Committed use discounts
Google Cloud's commitment-based discount: agree to a steady amount of usage for 1 or 3 years in exchange for a lower rate. The GCP analog to AWS Savings Plans and Reserved Instances.
Data transfer cost
What clouds charge to move data — out to the internet (egress), between availability zones, or across regions. It's often a surprisingly large, poorly-attributed slice of the bill because it isn't tied to any single resource.
DynamoDB on-demand
A DynamoDB billing mode that charges per request instead of for provisioned throughput. For spiky or unpredictable traffic it's usually cheaper than provisioned capacity, and it removes the guesswork of sizing read/write units.
EBS gp3
The current-generation general-purpose EBS volume type. It's cheaper per GB than the older gp2 and lets you set throughput and IOPS independently, so most gp2 volumes can move to gp3 in place — with no downtime — for an immediate ~20% storage saving.
FinOps
A practice that brings engineering, finance, and product together to manage cloud spend as a shared, data-driven responsibility — treating cost as an engineering metric like latency or uptime.
Idle resources
Provisioned resources doing little or no work — a database with no connections, an instance at 1% CPU, a load balancer with no traffic. Pure waste until they are stopped or removed.
Infrastructure as code
Defining cloud infrastructure in version-controlled files (Terraform, CloudFormation, and others) instead of clicking in a console. It makes optimizations safe and fast to apply: the fix is a reviewed diff, with an audit trail and a rollback.
NAT gateway cost
NAT gateways charge both per-hour and per-GB processed, so on network-heavy AWS accounts they quietly become one of the largest line items — often reducible with VPC endpoints or architecture changes.
Orphaned resources
Resources left behind after whatever used them is gone — unattached EBS volumes, unused elastic IPs, stale snapshots. They keep billing with nothing pointing at them.
Provisioned IOPS
Storage performance (input/output operations per second) you pay to reserve, e.g. on io1/io2 EBS volumes or RDS. Powerful for latency-sensitive databases, but easy to over-provision and keep paying for headroom you never use.
Realized savings
Savings that have actually shown up on the bill, confirmed by re-measuring the resource after a change — as opposed to an estimate. Lizrd tracks realized savings so a recommendation is only 'done' once the money is proven.
Reserved Instances
A 1- or 3-year billing commitment to a specific instance type in exchange for a large discount over on-demand pricing. Best for steady, predictable workloads.
Rightsizing
Matching a cloud resource's size — CPU, memory, instance type — to what it actually uses, so you stop paying for capacity that sits idle. It's one of the highest-impact, lowest-risk cost optimizations.
S3 Intelligent-Tiering
An S3 storage class that automatically moves objects between access tiers based on how often they are read, so cold data lands on cheaper storage without hand-written lifecycle rules. Well suited to data whose access pattern you cannot predict.
Savings Plans
A flexible AWS commitment to a steady dollar-per-hour of compute spend (1 or 3 years) for a discount that applies across instance families and regions — more flexible than Reserved Instances.
Showback & chargeback
Attributing cloud cost back to the team, product, or environment that incurred it — showback reports it, chargeback bills it. It is what turns a single opaque bill into cost that owners can actually act on.
Spot Instances
Spare cloud capacity sold at a steep discount that can be reclaimed with short notice. Ideal for fault-tolerant, interruptible workloads like batch jobs and CI.
Stop reading, start saving
Connect read-only and Lizrd finds these in your own cloud — with the exact fix.