Glossary
Definitions of Infra resource, placement, storage, and operation terms.
Infra separates ownership, placement, observed state, and recovery into distinct resource concepts. Use these definitions to interpret console details and API responses.
Terms
Agent
Node service that reports observations and applies authorized local changes. See Enrollment.
Allocatable capacity
Capacity available for admission after host reserve and policy. See Capacity.
Blueprint
Versioned definition and identity mapping for a group of managed resources. See Blueprints.
Cold move
Transfer of a stopped droplet when live migration is unsupported. See Lifecycle.
Connector
Authorization and reconciliation path from a self-hosted installation to hosted capacity. See Connector.
Control plane
Services that authorize, store, schedule, and coordinate management changes. See Architecture.
Cordon
Disable new compute placements on a node without stopping existing guests. See Node maintenance.
Desired state
Configuration accepted and stored by the control plane. See Resource model.
Drain
Evacuate node compute workloads through tracked child operations. See Node maintenance.
Droplet
Virtual machine owned by a project and placed in a region. See Droplets.
ETag
HTTP representation of a resource's configuration revision. See API.
Failure domain
Infrastructure boundary whose loss can affect colocated resources. See Regions and zones.
Fencing
Preventing a previous workload or storage writer from continuing before a replacement starts. See Recovery.
Guest agent
Optional in-guest service for observations and guest-assisted operations. See Guest access.
Headroom
Allocatable capacity remaining beyond reservations. See Capacity.
Health
Named conditions describing whether a resource operates as expected. See Metrics.
Idempotency key
Request identity used to recover the original result after a retried mutation. See API.
Image
Bootable guest disk artifact with immutable revisions. See Images.
Journal
Timestamped component or operation event records. See Logs and journals.
Live migration
Transfer of a running droplet between compatible nodes within one region. See Live migration.
Mesh
Regional node transport beneath isolated project networks. See Mesh networking.
Node
Registered Linux host providing compute, storage, or both. See Nodes.
Node pool
Group of compatible hardware and admission policy. See Nodes.
Observed state
Most recent state reported by an execution or observation component. See Resource model.
Operation
Durable execution record for an accepted infrastructure change. See Operations.
Overcommit
Admission ratio permitting reservations above eligible physical capacity. See Capacity.
Placement group
Required or preferred affinity or spread policy for droplets. See Scheduling.
Project
Ownership and access boundary for application resources. See Projects.
Quota
Admission limit on resource counts or configured capacity. See Quotas.
Reconciliation
Comparing desired and authoritative observed state to resolve incomplete work. See Operations.
Recovery point
Completed captured state available for restoration. See Backups.
Region
Compute, storage, and network placement boundary. See Regions and zones.
Replica
Live copy of storage blocks under a storage policy. See Storage pools.
Reservation
Capacity or quota held by a resource or accepted operation. See Capacity.
Revision
Configuration version used to detect concurrent updates. See API.
Snapshot
Volume block-state capture retained in its storage system. See Snapshots.
Spot
Preemptible compute capacity that can be reclaimed. See Spot droplets.
Storage policy
Pool, replica, failure-domain, and feature requirements for a volume. See Storage pools.
Taint
Node admission restriction requiring a matching toleration. See Scheduling.
Template
Reusable, revisioned droplet configuration pinned to an image. See Images.
Volume
Persistent block device with a regional storage policy. See Volumes.
Zone
Physical failure domain within a region. See Regions and zones.
Recovery measurements
Recovery point objective is the maximum data-loss interval an application plans to tolerate. Compare it with the capture time of the latest completed, usable recovery point.
Recovery time objective is the duration an application plans to tolerate before service restoration. Measure it through infrastructure restore, application validation, and traffic cutover during a recovery exercise.
Configured schedules and estimates do not establish either outcome. Use Disaster recovery to test them.