specification: API Commons Rate Limits
specificationVersion: '0.1'
schema: https://raw.githubusercontent.com/api-evangelist/interface-research/main/schema/api-commons.yml#/$defs/RateLimits
provider: OpenStack
providerId: openstack
created: '2026-05-04'
# Provenance stamped 2026-08-11: this artifact was written by the API Evangelist
# bulk sweep dated 2026-05-04, not harvested from the provider. See roadmap#35.
method: generated
modified: '2026-05-05'
reconciled: false
tags:
  - Cloud Platform
  - Infrastructure as a Service
  - Open Source
  - Rate Limiting
notes: OpenStack is self-hosted; rate limits are configured per-cloud by the operator (api-paste.ini /
  Nova/Neutron policy + quota configuration). Upstream documentation describes the mechanism but not numeric
  defaults that apply across all clouds.
description: OpenStack does not enforce a global rate limit. Each operator configures per-tenant quotas
  (Nova, Cinder, Neutron) and optional API rate limits via the api-paste.ini middleware (e.g. RateLimit
  filter). Public-cloud OpenStack providers publish their own per-account API rate limits in their respective
  documentation.
sources:
  - https://docs.openstack.org/
  - https://docs.openstack.org/api-quick-start/
responseCodes:
  throttled: 429
  overLimit: 413
  forbidden: 403
limits:
  - name: Per-tenant quotas (Nova / Neutron / Cinder)
    scope: tenant
    metric: varies
    limit: see operator-configured quotas
    notes: Quotas bound resource counts (instances, volumes, networks), not request frequency.
  - name: API rate limits (operator-configured)
    scope: tenant
    metric: varies
    limit: see api-paste.ini RateLimit middleware
    notes: Optional middleware; defaults vary by deployment.
policies:
  - name: Operator-controlled
    description: All rate limiting and quota enforcement is set by the cloud operator. Public OpenStack
      cloud providers (OVH, Rackspace, etc.) publish their own values.
  - name: Backoff
    description: Honor 429 / 413 (Over Limit) responses; exponential backoff with jitter is the recommended
      client behavior.
  - name: Quota vs rate
    description: OpenStack quotas (instance count, vCPU count) are admission controls; rate limits are
      a separate middleware concern.
maintainers:
  - FN: Kin Lane
    email: kin@apievangelist.com