Cloud pricing glossary
What cloud pricing terms mean on a real bill
This page is an index, not thirty compressed definitions. Each term has its own question-shaped page, a specific expensive mistake, a worked example, and primary sources. The examples use the provider prices checked on the date below rather than presenting them as timeless prices.
Example prices checked:
A
- What is an always-on minimum? An always-on minimum is the smallest continuously allocated instance, replica, or spend floor that remains even when user traffic is zero.
B
- What is a burstable vCPU? A burstable vCPU can run above a baseline for limited periods by spending accumulated CPU credits or by entering a paid surplus mode.
- What is block storage? Block storage presents a disk-like volume that a machine can format and mount.
- What is BYOC (bring your own cloud)? BYOC means the management platform runs workloads in a cloud account the customer owns and pays for separately.
C
- What are CPU credits? A CPU credit represents a fixed amount of full-vCPU work that a burstable instance can spend above its baseline.
- What is a control plane fee? A control plane fee is the recurring charge for the managed components that run and coordinate a cluster, separate from the workloads they schedule.
- What is a cold start? A cold start is the extra startup work and delay required when a platform must create a fresh runtime instance before serving work.
- What are committed-use discounts and reserved instances? Committed-use discounts and reserved instances exchange a time or spend commitment for lower prices on matching usage, and some reservations can also provide capacity benefits.
D
- What is data egress? Data egress is data leaving a provider or service boundary for a destination that the provider prices as outbound transfer.
- What is data ingress? Data ingress is data entering a provider or service boundary.
- What is a dedicated vCPU? A dedicated vCPU is processor capacity allocated for the sustained use of one customer's virtual machine rather than shared as a general-purpose time slice.
E
- What is an egress overage? Egress overage is the charge or policy consequence that starts after included outbound transfer is used.
- What is ephemeral storage? Ephemeral storage is tied to the lifecycle guarantees of its compute resource and is not a durable copy of data.
I
- What is inter-AZ transfer? Inter-AZ transfer is traffic between availability zones in one cloud region.
- What is inter-region transfer? Inter-region transfer is traffic that crosses from one cloud region to another.
- What is included data transfer? Included data transfer is the outbound allowance bundled with a plan before an overage rule applies.
L
- What does a load balancer cost? Load balancer cost is the recurring and usage-based price of the managed network service that distributes traffic across backends.
- What do list, calculated, usage-based and quote prices mean? We label every price with one of four types: a list price the provider prints, a price we calculate from the provider's own rates, a price that depends on your usage, and a quote you have to ask for.
M
- What is a minimum billable duration? Minimum billable duration is the shortest amount of runtime a provider charges for each start, even when the work finishes sooner.
O
- What is object storage? Object storage keeps named objects with metadata inside buckets rather than presenting a mountable block device by default.
P
- What is pooled data transfer? Pooled data transfer combines allowances and usage across a provider-defined account, team, project, or product group.
- What is a public IPv4 charge? Public IPv4 is a routable address that some providers now price as a separate resource even while it is attached and in use.
- What is the difference between promotional and renewal pricing? Promotional pricing is a temporary lower rate, while renewal pricing is the rate that applies after the promotion ends or the first term renews.
- What is a platform fee? A platform fee is the base charge for the deployment or management layer before separately metered infrastructure and usage are added.
- What is per-seat pricing? Per-seat pricing charges for each billable user or role that can access a platform.
S
- What is a shared vCPU? A shared vCPU is a virtual CPU whose underlying processor time is shared with other tenants and is not guaranteed to be continuously available to one workload.
- What is a storage overage? Storage overage is the charge for persistent capacity beyond what a plan includes, or for all separately provisioned storage when none is included.
- What is scale to zero? Scale to zero means a service can reduce its running compute instances to none when there is no work.
U
- What is unmetered transfer? Unmetered transfer means the provider does not charge each transferred GB under the stated policy, not that every amount can move at full speed forever.
W
- What is a worker node? A worker node is the compute machine that runs application pods and cluster agents under the control plane.